All right. Good morning, everybody. Thanks for joining us for the 2024 Thomson Reuters Investor Day. I think I know most of you. I'm Gary Bisbee, Head of Investor Relations. So when we began planning for this Investor Day, I told our leadership team I have four goals for the day. First, to explain our businesses and market opportunities to help build your confidence in our ability to deliver the revenue growth acceleration we called for on 8th February . Second, to reinforce the high quality of our key franchises. Third, to help you understand this is a changed company, primed for acceleration. And fourth, to showcase the depth and diversity of our leadership talent at these three tables here, which is exceptional. So we're really excited to have you here and to tell our story.
I hope we achieve all four of those goals, because if we do, I think you'll share our optimism and confidence in the positive prospects we see for Thomson Reuters. So how are we going to do this? Well, the agenda shown up on the screen here today, we have a series of presentations by key leaders of our company, which will be kicked off by Steve Hasker, CEO, in a minute. Our Chief Product Officer, David Wong, will host a product demonstration with six of his leaders showcasing some of our most innovative offerings. We'll have two Q&A sessions and a 15-minute break at 10:15 A.M. this morning. For those in person, we also, I think, as you saw, have five product kiosks outside the ballroom here.
They'll be manned during the break and after the end of our presentations to help you better understand the value that our products provide for our customers. Finally, we'll conclude with a lunch back here in the room for those who'd like to spend more time with our leaders. Before I turn it over to Steve, let me first cover the regulatory disclosure statements. Today's presentation contains forward-looking statements and non-IFRS financial measures. Actual results may differ materially due to a number of risks and uncertainties discussed in reports and filings we provide to regulatory agencies. You can access these documents on our website or by contacting Investor Relations. As Steve gets ready to head up the stage here, we're going to start off by playing you a short video that highlights a refresh of the Thomson Reuters brand that's being launched today in connection with Investor Day.
In times of change, even knowledge is a moving target. In times of complexity, professionals yearn for clarity. Professionals need the power to add greater value now, reduce inefficiencies now, anticipate future challenges now. The future is arriving faster than ever, but Thomson Reuters is already here, clarifying the complex with trusted expertise, powerful technology, advanced AI, and industry-leading insights. We're bringing professionals to a new era, a new time for Thomson Reuters, helping our customers know today and navigate tomorrow.
Please welcome Steve Hasker.
Good morning, everyone. Thanks to all of you for joining us today, both here in New York and on the webcast. Before I open up, I'd like to say a special thanks to David Thomson, our chairman, who's with us in person, Kirk Koenigsbauer, Board Director at Thomson Reuters, and Wolf von Schumann, a long-time Board Director. Thank you to the three of you for taking the time to join us. Let me briefly comment on the video you just saw, which brings a more focused and, we think, refreshed brand promise. Thomson Reuters clarifies the complex. We empower professionals to take action with greater confidence. With that brand promise clarifying the complex, that'll be a big theme today as we go through the presentations.
And as Gary said, we have a very full agenda to share with you, which we hope you will find both compelling and exciting. Our progress since the last Investor Day in 2021 has been substantial. We hope you agree with that. And it provides us with meaningful foundations from which to deliver consistent and improved performance in the future. To this end, our recent Q4 earnings call, during which we introduced a new financial framework with accelerating revenue growth. And today we'll go deeper into the key drivers of this framework with the goal of helping you better understand the factors that are contributing to our improving growth prospects. This will include a heavy focus on product, where we're really proud of what our teams have achieved in the last 18 months and how they're accelerating our innovation activities.
Our strong product portfolio and exciting roadmap, along with our go-to-market execution, will be important drivers of our future revenue acceleration. You'll also hear from the talented and capable team that we've assembled that's driving our execution. Many of the faces will be familiar to you. We've also had some new faces reflecting our continued efforts to build the strongest team in business information services. I think we've made great strides against this goal, and I'd confidently put our team up against any competitor any day. With that introduction, let's begin. I'll start by highlighting four key messages that I hope you'll take away from my words this morning and our Investor Day more broadly. The first message is that we have delivered against our 2021 targets while becoming a stronger company.
As I'll detail shortly, we have met or exceeded the 2021 to 2023 revenue, margin, and cash flow targets. More importantly, our progress has been significant at improving our company since our last Investor Day. This leaves us with much stronger foundations that we think support our outlook for faster growth. The second message this morning is that we operate in large and growing end markets with positive demand drivers. I'll discuss our addressable markets and growth momentarily. We see two market dynamics that are providing what we expect to be long-term tailwinds for our business. The first is the rising complexity of regulatory compliance, which is driving increasing need for trusted, accurate, and actionable insights in technology, which we are uniquely positioned to provide.
And the second is the potential of generative AI, which we believe positions us to play a larger role in the success of our customers as GenAI transforms the way that professionals work. The third message is that we are uniquely positioned to capitalize on these market opportunities. There's a few reasons for this: the significant competitive advantages we bring to our markets, our growing product momentum, and the optionality that our robust capital capacity provides us. The fourth and final message is we are investing heavily in 2024, and we're confident in growth acceleration in 2025 and 2026. I'll review our organic and inorganic investments, which provide confidence in our outlook for organic revenue growth acceleration. The legal, tax, and risks markets in which we operate are attractive, and they're growing. We hold leading market positions in each.
Demand for content-enabled technology is rising, and we see strong potential to drive innovation, including through generative AI. We have an attractive business model featuring highly recurring revenue and high customer retention, and a cost structure that supports healthy operating leverage over time. We are truly committed to following a disciplined financial approach that focuses on driving shareholder returns. It's a model that we do not take for granted, and we continue to work to strengthen and improve each of our positions every day. Our attractive business model is grounded in the strength of our big three franchises: legal, tax, and corporates, which leverage proprietary content, subject matter expertise, industry-leading AI and machine learning, and deeply embedded software.
The trusted, authoritative content and insights we produce, along with workflow automation and decision tools, are highly valued by our customers and would be extremely difficult to replicate, providing us with an important competitive differentiation. Let me also provide a few words on both Global Print and Reuters. Our Global Print business leverages the authoritative content from our legal and tax and accounting businesses and provides additional value to TR. Reuters News is a critical source of intelligence and insight for the professionals we serve within and beyond the big three customer segments, especially during times of change. Reuters is a powerful brand built on 170 years of independent, fact-based reporting overseen by the Trust Principles.
The generative AI advancements being made within our big three segments will be leveraged to both transform the productivity and the output of our newsrooms and also strengthen and diversify the Reuters business model by enhancing our news information offerings to customers like LSEG, media companies, and professionals. Let me now briefly review our progress since the last Investor Day in 2021. In March 2021, we discussed two key initiatives driving the change program, which were to transition from a holding company to an operating company and from a content provider to a content-driven technology company. We also discussed four key focus areas for driving improved performance for both our customers and our shareholders. While there remains work to do, we're very pleased with the progress we've made across both initiatives and all four of these focus areas.
Our transition to an operating company has yielded efficiencies, improved our ability to prioritize, and resulted in quicker decision-making. Our efforts to streamline our infrastructure, modernize technology, and invest more heavily in organic and M&A-driven innovation position us as a more tech-forward business, providing improved value to our customers. I'll mention a few more examples here. Under the first line, "Reimagine the Customer Experience" and the key focus areas, we have introduced digital customer service, improved customer support and call center answer times, and driven a more than 35% increase in customer NPS scores. Under "Optimize Products and Portfolio," we have divested or retired approximately 60 non-core products and made more than $2.2 billion of strategic acquisitions. We are positioned to invest in a rising proportion of our capital investments behind our best growth opportunities.
On the simplification and modernization front, we have reduced our office footprint by more than half and largely moved off of our legacy data centers and into the cloud. Then lastly, we have significantly upgraded our team over time while also leveraging our global centers to access new pools of talent in a cost-effective manner. Through these efforts, we believe Thomson Reuters is a changed company. It's worth noting that culture has played an important role in the change that we've delivered. Our focus is squarely on serving our customers. We don't fly in private jets, and we don't have extravagant executive offices. The hallmark of this team in front of you is that we roll up our sleeves every day, and we work as a group to identify and solve problems for our customers.
Compared to the TR of a few years ago, we believe we are more focused and performance-driven, with improved growth prospects and stronger profitability and cash flow. While the financial success of tripling revenue and increasing margins has been rewarding, I believe the long-lasting legacy of the change program and our 2021 to 2023 efforts will be the foundation they provide for improved, sustainable, profitable growth. With the momentum of our product and engineering organizations, we are increasingly confident in our ability to organically innovate and deliver improving, sustainable growth into the future. During this period of change and progress, we consistently delivered, as I said, against the three-year targets provided in 2021, achieving or exceeding our outlooks for revenue, growth, profitability, and free cash flow in each of the three years.
We have deliberately chosen to invest above our initial capital intensity targets due to our conviction as a team around the growth opportunities in front of us. This higher investment has largely been aimed at generative AI and also investing in several strategic acquisitions, which are expected to be key drivers of revenue acceleration in 2025 and 2026. Another key area of success for us has been our efforts to invest our capital capacity to create shareholder value. I think, as many of you know, since 2021, we've monetised $7.6 billion of our stake in the London Stock Exchange Group. This includes nearly $5.5 billion monetised in 2023 and another $1.1 billion sold just last week.
We've reinvested much of these proceeds and our free cash flow across a balanced and thoughtful capital allocation approach, which has included, of course, annual dividend growth, very selective strategic M&A, and returning capital through share repurchases, and, of course, the 2023 return of capital transaction. On the M&A front, we've been disciplined in our pursuit of strategic M&A since 2021 while also divesting a majority stake in LSEG and a number of smaller non-core assets. And as a result, our portfolio today is stronger, more strategically aligned, and with better growth prospects than two years ago. Looking forward, we remain focused on continuing this balanced and disciplined capital allocation track record. So while we're proud of what we've accomplished in the last few years, our growing conviction, and really the conversation today, is around our future potential.
I'll focus my remaining remarks on discussing our large market opportunity, the dynamics that are benefiting our growth, and why we believe we are poised for success with a bit more color on where we're investing in 2024. As we discussed in the past, we compete in large and growing markets. Within our big three segments, we estimate our vended market opportunity at $26 billion and our total addressable market, including generative AI potential, to be $84 billion. While there are a range of growth profiles within our market segments and our product categories, we see 7%-10% growth potential across our markets. Our big three segment presidents will discuss the TAMs and the growth in their respective markets in more detail later this morning. I'll summarize by stating that we see significant growth and market opportunity in each big three segment.
We believe that all three can maintain or accelerate that growth profile in the years ahead. I'd ask you to please note that the generative AI TAM uplift that we show here is based on our current roadmap and our current price points. We have a growing conviction that there could be upside driven by the opportunity to play a larger role in the success of our customers as generative AI automates and transforms the way that professionals work. It's too early to assess the end state and the precise timing of this, but our confidence in our ability to support our customers has never been higher and is supported by our daily interactions with those customers. Let me transition now to a discussion of two market dynamics we believe will provide meaningful demand tailwind for our businesses over the next few years.
I'll start with the rising complexity of regulation and compliance in our legal, tax, and risk-related markets. We share a few examples on this slide, which highlights the growing number of U.S. federal regulations, the rising volume and complexity of U.S. tax returns, and the steep increase in the number of e-invoicing mandates coming online over the next few years. The burden on corporations and institutions to comply with the growing patchwork of ever-changing regulations is very significant, as are the costs of non-compliance. Adding headcount to solve for this increasing complexity is not a sustainable nor a scalable option for corporations or their advisors. As a result, we see a growing and significant need for trusted, content-enabled technology, which is what we do best at TR.
We have a proven ability to capture, curate, and make sense of disparate datasets to then layer in AI-driven analytics and deliver key insights through modern, embedded software. Our content-enabled technology offers mission-critical solutions and insights that cut through this complexity, allowing customers to make timely and informed decisions. We believe we are uniquely positioned to help corporations and institutions navigate this rise in regulatory complexity, and we have a growing conviction that it will provide a long-term demand tailwind for our businesses. And as we discussed throughout 2023, we're also excited about the potential for generative AI to transform the work of professionals, and we believe we are well-positioned to be a key player in driving this transformation. Kirsty Roth and David Wong will review our GenAI infrastructure and our product strategy in more detail later this morning. But let me provide a few thoughts.
We see GenAI as a big opportunity, as I said, with applications across legal, tax, risk, and news markets. We already see this technology driving deeper integration between content and workflow software, which allows us to play a larger role in the success of our customers and, in doing so, expand our TAM. We bring significant advantages that position us to lead in the delivery of GenAI applications to our markets, and our conviction in that is growing. This includes authoritative and trusted proprietary content, extensive subject matter expertise, world-class data science and AI talent, 30 years of AI and machine learning innovation, and very significant distribution scale. We're investing heavily, as you know, and we're focused on a build-partner-buy approach. We've committed to spend more than $100 million annually in addition to inorganic investments such as Casetext, which we're extraordinarily excited about.
While it remains early, customer feedback on our strategy, our products, and our roadmap is extremely positive. The Westlaw Precision AI launch has driven meaningful sales momentum. CoCounsel is going very well, and the excitement about the other key 2024 launches is building, and you'll hear our product executives talk through some of those this morning. We're also very optimistic about the future potential to leverage generative AI ourselves internally to drive efficiencies and to make our colleagues' work easier and more fun. And Kirsty will comment on this a bit further in a few moments. So our excitement around all of this is bolstered by significant competitive advantages that we believe uniquely position us to deliver against this attractive market leadership. So let me just quickly tick through these. We have market leadership. We hold the number one or number two positions in most markets in which we compete.
We bring proprietary content and insights. Our content and datasets are unique and proprietary across multiple key franchises that you all know of: Westlaw, Practical Law, Checkpoint, CLEAR, Domínio, amongst others. Thirdly, we deliver mission-critical solutions. Our content-enabled technology is deeply embedded in customer workflows and essential to our customers' success, and we're very, very proud of that. We have built trusted customer relationships. We do not take them for granted. We have served our key legal, tax, and risk markets for decades with best-in-class offerings, enabling us to bring deep relationships with customers within those, including all of the Am Law 100, the Fortune 100, and the top 100 US CPA firms.
And then lastly, second last, we've built world-class talent, a deep bench of talent, which we think is an ideal blend of tenured TR colleagues with deep-end market expertise, blended with recent additions who bring new perspectives and experiences to our company. And then last but not least, we have leading scale and distribution. The depth of our offerings and the breadth of those customer relationships provides important distribution advantages, both in capitalizing on our organic product roadmap and also, importantly, integrating and growing strategic acquisitions. Now, in addition, we have growing product momentum that we expect to sustain over the next few years. You'll hear about this. It's a big focus of today. It's a big focus of this morning. We have a number of key products that are double-digit growers, as illustrated on this slide.
These offerings make up approximately 20% of our revenue and have contributed meaningfully to our growth acceleration in recent years. This group of products includes several that have compounded at high rates for a decade or more, including Practical Law, Domínio, and CLEAR. Others are more recent additions to the TR portfolio, including SurePrep, CoCounsel, and Pagero. Looking forward, we remain very confident in this group of offerings, which remain positioned to compound growth well into the future. And in addition, Westlaw, which makes up roughly one-quarter of total company revenue, has seen its organic growth rate improve meaningfully in recent years, boosted by both the Precision launch and, more recently, the strong sales of Westlaw AI. And we see this Westlaw growth continuing at rates above the historical trend through at least 2025.
So we're very excited about our robust product roadmap for generative AI offerings in 2024, which we expect to drive growing sales as we move through this year and especially into 2025. And we've been very focused on becoming the most innovative company in innovation and business information services, and we think we've taken some steps to achieving that goal. Lastly, we continue to have significant financial capacity, as I mentioned, which provides optionality to pursue both strategic M&A or other means to drive shareholder value. So after funding the Pagero acquisition and completing our ongoing $1 billion share repurchase program, we estimate our capital capacity through 2026 at roughly $8 billion. Mike will discuss in some detail our disciplined, returns-based approach to M&A in his presentation later this morning.
I'll just reiterate that we remain focused on executing strategic M&A and also bolstering our big three businesses while executing our proven acquisition playbook of buying pristine products and putting them through our distribution. In closing, a few thoughts on our 2024 investments and our forward outlook. In summary, our conviction around the medium-term growth potential for Thomson Reuters is rising, bolstered by several of the items I've discussed today, including healthy end-markets, complexity tailwinds, generative AI potential, our own product momentum, and the benefits of our 2023 to 2024 portfolio of M&A. Given this attractive backdrop and the many opportunities in front of us, we've stepped up the organic investment through 2023 around generative AI and through M&A, and we're accelerating that investment further in 2024.
We think we do this from a position of real strength, and we're very bullish on the returns the investments that I've described will earn, and we expect them to deliver in the form of revenue growth acceleration and return to year-over-year margin expansion in 2025 and beyond. With our fourth quarter results last month on February 8th, we provided a financial framework for 2025 and 2026, in which we see organic revenue growth accelerating to 6.5%-8% from approximately 6% in 2024. We're very focused on continuing our track record of strong execution, and we have confidence in our ability to deliver to this outlook, which benefits from both recent M&A and the execution of our generative AI and other investments.
So in conclusion, we've delivered on our commitments. We operate in attractive-end markets with several very important tailwinds. We are uniquely positioned to capitalize on the opportunities that I've described that we think are in front of us, and we're equally confident that our investments will deliver accelerating revenue growth over the next few years. With that, let me now introduce Kirsty Roth, our Chief Operations and Technology Officer. Thanks, everyone.
Thanks, Steve. I'm Kirsty Roth, our Chief Operations and Technology Officer, and I'm accountable for delivering on the plans from product and our segments and providing a great experience for our customers by innovating faster with engineering and using our operational teams to provide accurate and timely content, as well as ensuring first-class customer support. I have a track record of execution, most recently demonstrated by the execution of our change program. Let me start by highlighting the three key messages that you'll hear from my presentation. First, we've delivered against our change program initiatives, making significant progress in areas like product stability, engineering velocity, and call center quality. Second, I will discuss our areas of focus in customer service for the next three years.
Then finally, I will discuss our generative AI technology platform, which is increasing the speed to market of our new generative AI features while ensuring quality and security. As you've already heard from Steve, we've achieved a lot over the past couple of years. When we spoke to you in 2021, we highlighted the five priorities you see on this slide, and we have made significant progress in each of these. Today, our customers get to enjoy improved performance and stability of our strategic products, with 97% of them meeting our performance targets, which include a 99.5% or better availability metric. A simplified product portfolio that we have rationalized from approximately 170 to around 110 products through a combination of platform rationalization and exits, enabling much better focus for our product and engineering teams.
We've increased our velocity with updates to over 85% of our products now released every single week, when previously, the majority were released quarterly. We have matured our API strategy, with our gateway now receiving over 1.5 billion calls monthly and invested heavily in creating shared capabilities that allow us to reuse key functionality in our products. To this end, we've released around 20 shared capabilities, the most popular of which with customers is single logon. We now have over 90% of our revenue cloud-enabled, giving our customers more choice and modern solutions. We've invested heavily in customer support that is faster and more effective than ever, as demonstrated by a much faster average speed of answer and a first call resolution rate that is now over 75%.
In addition, we have embedded self-help across a number of our strategic products, with over 6.5 million self-help sessions in 2023. Overall, our NPS has improved from 16 to 23. But do note, NPS is a lagging indicator, and we expect to see continued improvement in the coming years as we continue to focus on our user experience. Finally, we have also delivered on digital enablement, ending 2023 with approximately $300 million in digital sales and renewals, which includes features like AI-driven upsales. Westlaw, Practical Law, and FindLaw also leverage digital assist, which frees up our sales teams significantly. That said, despite material progress, we still have a lot we want to focus on to do better for our customers over the next three years. We will continue to improve NPS by focusing on stability and user experience, getting all of our products to our world-class reliability metrics.
Now that we have established the shared capabilities for our products, the next few years will be less about new capabilities and more about increasing adoption in our products. This will enable our customers to enjoy a more streamlined product experience, where our products not only look the same, but they feel the same as customers switch between them. We will continue to expand our suite of self-help capabilities powered by AI. More on that in a minute. As mentioned earlier, we are also doubling down on our digital strategy with the aim of expanding digital sales and renewals to new products and geographies. For example, we are implementing digital sales for CoCounsel, SurePrep, and other legal and tax products, while expanding the offerings to Canada, the UK, and Europe, driving the growth of our digital channels.
We've also been supporting our go-to-market teams with improved capabilities in marketing, pricing, and retention. We have planned substantive investments in these areas to lift net revenue retention, or NRR, over the coming years by launching these further across the big three. Lastly, we will continue our focus on productivity through various initiatives, ranging from building out new capabilities in our global centers to unlocking productivity with the internal use of generative AI, the latter of which I will also speak to a bit more in a moment. Our global centers have evolved substantially since 2020. When they were around 31% of our footprint, today they are where 45% of our employees are based, including many of our leaders. We added a new global center in Mexico in 2021 to better support time zone-related work and enable marketing and sales support.
This has been very successful, and we will continue to evolve our presence in line with business need. Finally, we intend to complete our cloud-enablement program this year, and then we will focus on the efficiency of our colocation data centers. A little more on our plans for customer support. We will continue to build upon our self-service and help center capabilities, leveraging AI wherever we can. Today, more than 80% of our products have help content available, more than 20 of which have this in-product support. We are seeing an all-time high in visitor numbers and self-help sessions. We have robust chat and chatbot experiences today, the former powered by humans and the latter by AI. Our customers are active across seven communities where they really help each other get the best out of our products.
Our overarching goal for self-help initiatives is to establish a personalized support journey for our customers, seamlessly integrating chat support, dynamic user communities, and robust content and training, driving improved NPS, faster issue resolution, and deeper product adoption. To this end, our three main priorities revolve around product-agnostic intelligent chatbots, onboarding of more products into self-help, while also increasing customer adoption of self-help itself. I'd like to call out our intelligent chatbot plans in particular. We are planning to deploy more product-agnostic intelligent chatbots powered by generative AI and our search capabilities. This will include leveraging our LLMs to help our customers directly, as well as assisting our support agents in answering customer queries faster and more accurately. In addition to using AI to improve our products, we have begun to use it internally to improve productivity, with some really promising early results.
We've established a focused program led by our Chief People Officer, Mary Alice Vuicic, and myself, supporting all of our teams across TR to make their jobs more enjoyable and productive. Our early progress has shown real wins in content, customer support, and engineering. We are now rolling these out more broadly, and we are also seeing promising early results in both marketing and sales. We have three types of use cases we are progressing. The easiest: use of Microsoft products like Copilot and GitHub, where we can leverage our partnership. Secondly: leveraging SaaS products we use, like Salesforce and Workday, where we have a CLEAR roadmap. And thirdly: using our own teams to build bespoke solutions. In engineering, we have seen a good productivity uplift, improving quality and efficiency of our written code, and good improvements in test automation.
In customer support, we have so far built bespoke solutions to achieve more accurate and timely answers for our customers in multiple channels and for our own call center teams to answer calls more effectively. In content, we have bespoke solutions showing good wins on document curation and metadata extraction, and an editorial on summary creation and indexing. This is a substantive program that could yield meaningful benefits over the long term. While it's premature to provide financial targets given the early stage of our work and that of third-party tool providers, we are highly optimistic that our internal generative AI efforts will both contribute to future margin expansion and fund additional growth investments.
Before handing over to David Wong, who's going to walk us through the plans for our products, I would like to spend a few moments talking about our generative AI platform, the enabler that lies at the core of our ability to design, develop, and launch new high-quality generative AI features at speed, securely. So what exactly is the generative AI platform? It's a common development platform to design, build, and deploy generative AI skills. It leverages reusable components as the building blocks for future products. Every time a developer uses one of these building blocks to access content, enable login, or train a model, it speeds up their efforts. Given the pace of change in this AI world, these time savings really matter.
A number of the change programme initiatives to modernize and upgrade our technology, including our move to the cloud and investments in APIs and shared components, provided the critical foundations for the generative AI platform. In 2023, our colleagues from Casetext and TR Labs have leveraged this foundation to move from concept to reality with this key generative AI platform. For those of you that want a bit more detail, let me highlight the four key components of our platform. Firstly, our content platform provides access to TR's vast content, preparing and refining it to optimize the build and runtime in our products. Secondly, the AI platform enables experimentation through production to a full range of services safely and securely and in line with our Trust Principles.
Thirdly, the shared capabilities support AI skills, content services, the AI assistant which David will describe later, and foundational capabilities like API access. And finally, the skills development framework, or AI Skills Factory as we refer to it, ensures we have fast and reusable skills that our engineers can deploy into our products. Let me close on the platform by reiterating the key takeaway from my perspective. We have invested in infrastructure that positions us to innovate around generative AI securely and with speed. We see this as a competitive advantage as we work to deliver against the product roadmap that David will share with you. As a proof point, we believe our launch last fall of generative AI capabilities in Westlaw Precision was accelerated by around 4-6 months due to the benefits of this technology platform. Let me close with a final thought.
We've set up the platform so that the engineering teams have access to multiple large language models, as models differ at what they are best at, for example, search versus summarization. We currently have multiple LLMs in production and trials ongoing with a number of others. We are also building our own small or specialised language models, which we believe can achieve better performance in certain situations. We believe these efforts position us well to succeed amid the inevitable advances that generative AI technology will take in the future. Thank you. Now I'd like to introduce David Wong, our Chief Product Officer.
Right. Thank you, Kirsty and Steve. My name is David Wong. I'm the Chief Product Officer here at Thomson Reuters. I've had the privilege of leading our product, design, and content teams for the past four years under Steve's leadership. And I could not be more excited to drive our generative AI product development. It's not without exaggeration that I feel that this is the most energizing and meaningful work that I've done in my career. And I've spent nearly 15 years building B2B products at the intersection of data, machine learning, and AI. And I believe I bring three things to this challenge.
Number one is I've experienced working with expert users and building software that works for them that respects their knowledge and expertise. This is just as true for expert researchers in law and accounting as my previous work with expert researchers in marketing. Number two, I bring a strong technical foundation and know-how to build with novel AI technology. At one point in my life, I was a half-decent engineer. And I even coded my first neural network when I was 16 years old on what is now a standard training problem for budding data scientists. So I have the scientific and engineering background to understand and inquire and ultimately apply this new technology to problems.
And number three, I bring a diverse experience working in engineering and product organizations, from legacy mainframe groups to cutting-edge big tech teams to bring a pragmatic approach on how to approach accelerating innovation. Today, I'm going to share my views on TR's product strategy, why we're positioned to win, and our vision for how generative AI will transform our customers' experience. So Thomson Reuters' product strategy builds on our purpose and ambition. At the end of the day, what we do for our customers through our products is that we help professionals do complex and substantive work, both more efficiently and better.
What we mean by more efficiently is that means with less time, with less cost, and with less effort and human drudgery. By better, we mean higher accuracy, with higher quality, and that which ultimately creates better results for those professionals' clients. In the long term, we think that better will also mean greater creativity thanks to the use of artificial intelligence. We think that Thomson Reuters is uniquely able to deliver on this product strategy thanks to two historically strong assets. First, proprietary content and resources for professionals, plus the experts and know-how to produce and advise on how to apply that content on professional work. Number two, as Steve mentioned, software tools which are deeply embedded in professional workflows, such as tax preparation, legal workflows, and corporate tax and compliance processes.
What is unique about this moment is that we have these two assets which can be both exploited by GenAI technology, which you see in the center of this slide. Thomson Reuters, we have worked with AI for decades, so we've always understood its importance. What we couldn't predict was the launch and the evolution of ChatGPT, GPT-4, and the current crop of cutting-edge large language models. Like everyone else in the tech space, we had to shift gears, and we had to do it very quickly to respond. I think, thankfully, unlike many of our competitors, we are uniquely positioned to take advantage of this AI revolution. We have the teams, as Kirsty mentioned, to build it. We have the content to train it. I'm going to spend a little bit more time speaking about that. We have customers who want it.
Our segment presidents will share more about those customers and the demand that we're seeing. One year after the launch of GPT-4, we are kicking off 2024 with not just one GenAI product feature, but a whole portfolio of solutions, which I'll share more about, and my colleagues will demo for you. On top of this, we're benefiting from the change program. Our content and technology systems have been migrated to the cloud and made more safe and resilient so we can deliver on the promise of being the most trusted provider of AI solutions. As Kirsty mentioned, we are already in the progress of steadily modernizing the design and the experience of our products, bringing those industry best practices to our product development teams. Let me take a moment to just take a closer look at how we support professionals.
While Thomson Reuters has a portfolio of about 110 products, when we boil it down, we generally support customers with three broad jobs to be done. Number one, information retrieval, finding the right answer to an often nuanced and complex question. Work product creation, which is creating written work products such as legal documents, tax returns, compliance documentation. And managing risk and supporting decision-making, such as whether to sign on a new customer, to use a vendor, do business with a new partner. There is a range of jobs in each of these categories, from simple tasks like finding factual information to highly complex and subtle research tasks on novel issues. What we're seeing is that our most recent products and features, especially those that use AI, are delivering on those more challenging jobs at the bottom of this slide.
So as Steve mentioned upfront, we believe that we're going to increase our role in our professional customers' day-to-day work and be an increasingly critical technology provider to their firms as we take on these more challenging problems. On top of this, Thomson Reuters has this unique content and know-how, which enables our AI ambitions. Many companies have data, and many have experts, but few have both in such large quantities and focused in a particular domain. And on top of this, we know that we're very early still in this journey to apply AI. And there's going to be substantial change, new methods, new techniques, new approaches to take advantage of it. What I can tell you today is that Thomson Reuters is advantaged in applying pretty much all the major ways of using GenAI at this current moment. Let me just spend a moment to talk about that.
On the left, if we want to train new models, if we want to create small purpose-built models, we have the content in large enough quantities to build those custom models and the know-how and expert staffing to fine-tune those models as well. On the right, if we choose to use generic models off the shelf in combination with our search and with our data assets, with techniques such as retrieval augmented generation or RAG, we have a state-of-the-art search algorithm. We have experts with practice experience to create workflow and prompts. So I believe that we're uniquely able to adapt to change, and we can keep pace with this fast-moving space. So speaking of fast-moving, Steve mentioned that Thomson Reuters has been accelerating its pace of innovation, and particularly for the past few years.
In the last four years, we have accelerated with a combination of technology and talent investments. As Kirsty mentioned, we've migrated more of our software to the cloud with API-based architecture, and that allows for faster deployment and more code reuse. We've organized our teams into cross-functional pods with improved balance between engineering, product, design, and content experts. And we've adopted a more iterative and agile development approach, which has increased the clock speed of our teams, with many teams shipping every week and a few even daily. And if you look at the product releases that we have made that primarily apply AI and machine learning, this change is even more stark. Take the launch of Westlaw AI-Assisted Research, for example. GPT-4 was made available to the marketplace in February 2023. We started development on Westlaw AI-Assisted Research in March.
We announced our plans to the marketplace in May. We launched a beta in August. We shipped to thousands of customers and tens of thousands of users in November of last year. In just over eight months, we launched one of the most transformative product experiences in Westlaw's history. This is a new Thomson Reuters, and we are just getting started. For the rest of 2024, we have an ambitious roadmap of growth opportunities. I'm not going to read every bullet on this slide. To highlight our big themes for this year, we are completing our integration of Casetext into Thomson Reuters. We are going to start delivering a vision of a single, consistent CoCounsel AI experience to our customers. We're expanding the research and knowledge capabilities of our flagship products, Westlaw, Practical Law, and Checkpoint. We're also going to start integrating more with third parties.
We hinted to this last year with Microsoft 365 and the Copilot work. We're going to extend this to document management systems and, on the tax space, ERPs like Oracle and SAP for a number of our tax solutions. We're also going to maintain the pace of experimentation. We have new proofs of concept, and we're going to take on increasingly challenging and complex task automation with AI this year. And outside of legal, we will bring the AI system concept to accounting, audit, tax, risk, and fraud in 2025 and beyond, building on those shared capabilities and the foundation we've laid in legal. I also want to emphasize that our opportunities are not only in AI. We see growth opportunities in transactional compliance with our recent Pagero acquisition, new tax and accounting automation needs, and new risk and fraud use cases.
So before handing off to my colleagues to demonstrate our products, I want to spend just a little bit more time on our long-term vision for how generative AI assistants will augment professional work. We believe that in three to five years, every professional will have a generative AI assistant. This will be an assistant that you can delegate substantive professional tasks to, just as you would to any person on your team. This is an assistant that is knowledgeable about the facts and the professional standards of the work that you do, that works at superhuman speed, and that delivers at or above human levels of accuracy and quality. What I'm showing here is our vision for the legal AI assistant. CoCounsel, as I mentioned before, is the name of that legal AI assistant.
You work with it, and you ask it for help like you would an associate on your team. It has skills. It has skills which are highlighted in the dark green boxes, such as essential legal tasks like drafting emails, creating timelines, searching through documents. It has content-driven skills like Westlaw and Practical Law research. It has more sophisticated document analysis skills, and it helps you to draft. And as we create more skills powered by that generative AI platform and the Skills Factory, and as we improve our content and use increasingly more sophisticated LLMs that are going to be released by the different AI vendors, this assistant will become more valuable and more powerful. Again, all of this is built on top of the GenAI platform that Kirsty teed up for us.
That platform also means that we can take the same approach, and we can use the same infrastructure to deliver for the other professional segments that we serve, like tax, accounting, audit, and risk. I often get the question from customers, "This is great, but what will it look like? This is just a slide with some boxes, especially as it pertains to our integration of Casetext and CoCounsel into Thomson Reuters." So let me show you how this could look like for our customers in the future. In this scenario, I'm an associate at a law firm. I just finished a client meeting with a partner on an M&A matter, and I'm running back to the office. I get a message on Teams from the partner asking about some work that we can get started.
The meeting went well, and the partner wants me to start drafting an M&A agreement and to do better research. So what do I do? Well, again, CoCounsel is like a member of my team. So with CoCounsel integrated into Teams, I forward the message to CoCounsel just as I would to another member of my team. CoCounsel understands the context of the question, recognizes the forwarded message as a new request, starts a new chat, and also analyzes the thread, understands that we're doing this new acquisition for this company called Gamalinda. It's made up. And then it recommends two tasks for me to help with drafting and research in Practical Law. I didn't need to know which TR product to use, which content set to draw from. CoCounsel knows how best to help me.
I'm good with those suggestions, and I confirm, and CoCounsel gets to work while I finish my commute. I get back to the office, open up my laptop, and fire up Practical Law. And I see a notification. I see a notification that CoCounsel has completed the two tasks that I requested while I was in the car. Let me highlight a few things here. CoCounsel here is integrated right into Practical Law, and it works alongside you inside the TR product. This is a pattern that we're looking to take across all of the TR products. It also shows up when it's relevant to the user and is easily accessible. So when I click the CoCounsel icon in the left-hand bar, it opens a side panel. I'm a busy associate, so I might be working on many different matters, not just that M&A deal.
So I look at the conversation history with CoCounsel, and it pops the full list of matters that I might be working on. I click on that Gamalinda, Inc. chat that I started when I was in the car, and I can then see that those tasks are done. This is a really subtle but important point. CoCounsel, it has memory and shared context across our products and integrations. This means that if I start a thread, I can continue that thread across all the different products that we have. After clicking on Open in Practical Law on that research task, CoCounsel guides me to the resource that Practical Law has identified to support me. And right here, I've got a practice note for a merger agreement, which is Pro-Buyer. While reviewing that resource, I have a question.
I suddenly have a question on breakup fees, which might be problematic in this case. So I do some case research. Because CoCounsel has skills enabled by all our TR content, it knows the best way to answer the question, and that is to do research in Westlaw on that primary law content. CoCounsel runs the Westlaw research skill and, again, directs me to the right content and the right product without me having to remember where to find it. It brings forward an answer to that question in the chat. If I click one of the footnotes on one of the supporting materials, it takes me into that case within Westlaw. Again, didn't need to know where to find the resource. It just takes me there. Also, I have access to the full suite of all those Westlaw capabilities.
If I want to continue my research, I can then dig in, use products like KeyCite. I can use all the headnotes. I can use all the advanced research features inside Westlaw to continue and dig into my research. Because, as we know right now, these AI assistants are not good at completing the work. They give you a really good head start. And so we think that there's a strong connection between the interactive experience on the product and the chat. I'm satisfied with the answer. And so now I'm going to go back and take a look at that M&A agreement draft that was started. Again, I take a look at the results of those previous tasks that CoCounsel has started for me. I click, and I'm taken into Word. Again, in the sidebar, we have that Gamalinda, Inc. chat. We have the thread. We have the history.
The conversation travels with me to Word. No matter which product I'm in, I know that the context of my work is preserved. It also gives me access to all the drafting and the editing skills that I have inside Word, as well as from Thomson Reuters. I finish this work. I create the draft. I add in the provisions on breakup fee that I'm comfortable with. I'm done. Done the work. I get onto the train. While I'm on the train, I ask CoCounsel to help draft me an email to summarize the day of work, to be able to attach the draft agreement, and to be able to give an update to the partner. Because CoCounsel has that thread of work, it knows the context of the questions I've asked, the data that's come back, the content.
It can easily create the email which summarizes the state of work. I open that in Outlook. I review. It attaches the draft agreement and sends it to the partner. The partner gets it. So what I've shown you is work that would have normally taken days across half a dozen different TR and Microsoft products, done in a few hours thanks to the CoCounsel legal AI assistant. And to recap what is unique about this approach, the AI assistant is powered by a combination of cutting-edge LLMs, our content, and software solutions which are used by lawyers. And there's automation in all of those steps. Number two, it's available where the lawyer works. And it's as easy to use as talking to a colleague. It breaks down the silos and barriers between TR's products and solutions, which you think will provide a much better experience.
It's trusted and designed for the professional, supported by this skills-based approach. I hope you're as excited about this vision as we are. And again, we are just getting started. Now, I'm going to invite leaders from our team to demo a selection of our products. As you see on the screen, we are going to give a cross-section of products across legal and tax. We are showcasing these specific ones because they highlight the progress as well as the potential of AI and automation to drive growth into some of our biggest franchises. Each of our product leaders will illustrate the speed of our innovation, strong customer signal on demand, and the value of our product and expertise to create these solutions. With that, please welcome Mike Dahn, Head of Product for Westlaw, to the stage to kick us off. Thank you, David. Good morning.
My name is Mike Dahn. I lead our Westlaw product team. In November of last year, we launched the biggest change in legal research since the move from print research to online research. That was with the launch of AI-assisted research in Westlaw Precision. AI-assisted research uses the very latest in large language model technologies combined with Thomson Reuters' market-leading content and expertise. I'll leave you with two key messages today. One is that generative AI is going to be absolutely transformative for legal research. It already is, and it's going to be even more transformative in the future. Two, Thomson Reuters is uniquely positioned to provide the best generative AI solutions for legal research because of our market-leading content and expertise and the investments that we're making that we've made over decades but we're making right now as well.
Now, before I show you how this works, let me share some brief context to highlight what an important transformative change this is. When we talk with our customers, when we talk with lawyers about the work that they do and the legal research that they do, they tell us that for complex research, complex research makes up about a third, on average, of their research. And projects in that category tend to take about 12 hours, on average, across our customer segments. But for our largest law firm customers, it can take up to 22 hours, on average. But it's not just the time that it takes. When we talk with lawyers about doing this work, they'll tell us that even after 10 or 20 hours of legal research, they're still often worried that they've missed something important in their research. And why are they worried?
They're worried because research is critical, can be mission-critical for the work that they're doing. It can impact the outcome of cases. It can impact the quality of work that they do and the advice that they give to clients. They're worried, too, because mistakes actually happen. We hear about mistakes all the time talking with lawyers, talking with corporate counsel, talking or even reviewing published case law. We see that lawyers make mistakes. They miss important law in their legal research. So corporate clients really care about this. Excuse me. Corporate clients really care about this. They want to know that the advice that they're getting and the work that their law firms are doing is accurate and correct. They also care about efficiency. They want their lawyers, especially when they're being billed by the hour, to work fast and efficiently and be right.
That's why we've been investing so much for many decades to tie the law together so that lawyers can do legal research really quickly and do it really well with a high degree of accuracy. But even with all that we've done in legal research over the decades, what we see when customers come to Westlaw with a legal research question like, "Does a False Claims Act claim for retaliation include retaliation by the employer for the employee's conduct after termination?" This is the kind of question we get in Westlaw every day. And when a lawyer asks a question like this in Westlaw today, what they get back are long lists of cases, statutes, regulations, secondary sources, briefs, trial documents, expert witness testimony, etc. And then they have to read through all of that material trying to figure out the answer to the question.
If they miss anything important, they could be in trouble. With the launch of AI-assisted research in Westlaw, the process is so much better, so much faster. Now the lawyer can ask the question just like they would ask a colleague, just the way that they're thinking about it. When they hit search, our AI is examining those cases, those statutes, those regulations. What we come back with is not just a long list of cases and other materials. We're coming back with a narrative, responsive answer to the question that pulls together the right material from that law and pulls it together in a way that makes it really easy for the lawyer to understand what's going on. For instance, in this answer, we see that there's real nuance to the answer, that the law is different in different circuits.
We're explaining that to the lawyer. Now, the lawyer can't just use a response from AI when writing a brief for the court. We provide the actual language from the law in these footnotes. They can click on the footnotes to go and examine the key material, the cases, and statutes, and regs in more detail. This is a far better workflow than what they've done before, where they have to go through all of the cases, all of the statutes. Now we're pulling it together for them. We're moving from a situation where they're just collecting the law to actually providing a first draft of work product. This is the kind of response that they can use to start their brief or to start providing a memo to the client to provide them with advice. Now, there's more work for them to do.
As David mentioned, this is part of a process. We're combining it with the other technology in Westlaw so that they can dig deeper, find out what they need, and be sure that they're giving clients what they need. But they can do it now so much faster and better than they've done it before. It's not just about speed. We're also providing nuance in these answers that might have been missed in traditional research. Now, you've likely heard about an issue with the use of large language models called hallucinations. This is where large language models can fabricate or make up answers that are not true. That's because they're relying on their statistical understanding of language patterns to provide their answers.
So what we do with our solution in Westlaw is we use a process called retrieval augmented generation to prevent or largely reduce hallucinations with Large Language Models. So with retrieval augmented generation, we're essentially bringing the right material to the language model. And we're essentially saying, "Don't just rely on your statistical understanding of language patterns to provide this answer. Provide the answer based on these cases, these statutes, these regulations to provide your answer." Now, retrieval augmented generation is not a proprietary Thomson Reuters process. retrieval augmented generation is something that many companies will use to reduce hallucinations. But as soon as you realize that this is a necessary process for reducing hallucinations, then you realize that the quality of the discovery process and the quality of the content that we're bringing to the language model really matters.
The big problem to solve with the use of large language models is not hallucinations. That's a relatively easy problem to solve. For instance, we can have an answer come back from a language model that says, "The holdings of case X and the holdings of case Y address your question." Based on that, the answer would be yes. And that answer can be hallucination-free, meaning that there actually is a case X, and there actually is a case Y, and we could have the correct citations for those. But it could still be wrong because there may have been another case that overruled those cases, or there might be a statute that invalidates the holdings of those cases. The big problem is not preventing hallucinations. That's certainly a problem. But it's relatively easy to do that with retrieval augmented generation.
But because you have to use retrieval augmented generation, then you really have to focus on accuracy. Because for legal research, retrieval augmented generation is not necessarily enough by itself. You have to find the right law. You have to bring the large language model, the right law first, which means you have to find it. And then you have to get the language model to focus on the right parts of the right law. So you don't want it focusing on the context section of a case or the dissent of an opinion. You want it to focus on the precise holdings that matter. And then you've got to figure out how to resolve conflicts in the law. So there are some cases that say X and some cases that say Y and maybe a statute that says Z.
All of that has to be resolved when bringing this material to the language model if you're going to get an accurate response back. Fortunately, at Thomson Reuters, we've been working on these problems for decades, in some cases over 100 years. We've been tying the law together. We've been classifying it and organizing it so we can solve these problems. We can find the right material because we have it so well connected with the editorial work that we've been doing. We can get the large language model to focus on the right parts of the law because our attorney editors have been doing that for decades. We show very CLEARly, "These are the holdings of these cases. This other material is not part of the holdings." Then we can help resolve those conflicts where different laws say different things.
Because again, our attorney editors, with their work with KeyCite, have been figuring that out. They'll say, "This case overrules that case," or, "This statute invalidates that case." We can use that information to provide the right, accurate material for our customers. Now, as much work as we've been doing for decades and as we're doing right now, the system is still not perfect. The use of large language models still will provide inaccuracies sometimes. That's why right on the homepage of our product, we very CLEARly tell customers, and we tell them very explicitly in our training as well, that AI-assisted research uses large language models and can occasionally produce inaccuracies. It should always be used as part of a research process in connection with additional research to fully understand the nuance of the issues and further improve accuracy.
So this is something we're very CLEAR with customers about. They understand that using this is part of a process, part of a research process with all the other tools in Westlaw. We always tell our customers, "Use this to dramatically accelerate thorough research, but don't use it as a replacement for thorough research." They understand that intuitively right away. They told us that in the research that we did with them, they said, "If you're using AI, if I know that this is being produced by AI, I'm always going to be checking and looking for additional nuance."
That's why we put AI right in the title of the product name because we want them to be CLEAR about that. They appreciate it. With this knowledge and with this caveat, we talk to customers, and they tell us that it transforms their work, that it's a very big deal. It saves them hours and hours of time. It helps uncover nuance that they might have missed in traditional research. The term that we hear most often when talking with customers about this is game changer. They keep saying, "This is a game changer.
This is so much different from the work that I was doing before and helps me so much." When we talk with corporate counsel about this, they will say, "Wow, this answer, this is the type of thing that I would have hired a law firm to give me." When we talk with lawyers, they've looked at some of these answers and said, "This is what I would have come up with myself." They've run searches based on their own expertise. They look at the answer, and they know that that's what they would have told the client. They're really seeing that this is a big transformative change. We're not remotely done. As David said, we are just getting started.
There's so much more that can be done with generative AI to help our customers with legal research and with so much more that you'll hear about later. When customers get these answers back in Westlaw, we ask them, "What are you going to do next? What's your next step?" And they'll say, "Oh, I want to dig deeper. I want to understand the issues in more depth. I want to figure out if there's precise language that could be used to make my argument more compelling." And so we're going to help them with that additional research. We can use AI and agent-like technology to sort of emulate the best practices of great legal researchers and uncover a lot more detail for them.
So when they click a button like this, we can say, "Our AI has examined the law in much greater depth based on what you've given us so far. And here's all the nuance, all of the issues and the outcomes, the factual scenarios, the causes of action, the related exceptions and defenses on related concepts." And now they can tell us more about what they want. And we can ask them intelligent questions about the work that they're doing. So based on the answers to these questions, we can say, "Well, our AI has analyzed this law. And based on the information that you've given us, we can tell you that this law applies now or it doesn't apply," and help them focus and get to the end of their research much faster and get to better work product.
Now, every year for 20 years, the American Bar Association has been conducting an independent survey of the market. One of the questions they ask about is legal research. Every year for 20 years, the market has shown a strong preference for Westlaw. It's because we continue to invest. We continue to invest not only in our editorial work but also in technology. We've been using AI in Westlaw since 1993. We've continued to make it more and more sophisticated. You saw a big leap forward last year with now the use of large language models. That's only going to continue. You heard from both David and Steve that we're continuing to invest in a big way in this technology.
And we're also continuing to invest in our editorial resources so that we can provide even better solutions in legal research for our customers for years to come. So I'll leave you with this. Generative AI is a very big deal for legal research. And Thomson Reuters is uniquely positioned to provide the best generative AI solutions for legal research because of our market-leading expertise and editorial assets that we've built up over decades and our commitment to investment. We're investing a lot right now. And we're really excited about this work. As much as we've done and accomplished in the past few years, we're even more excited about the future. It's going to be great. And so with that, I'll turn it over to my colleague, Emily Colbert, to tell you about the big things that are coming to Practical Law.
Thanks, Mike. And hi, everyone. I'm Emily Colbert. I lead Practical Law product management. Building on what Mike just said about the importance of our content, of high-quality content with generative AI, the editorial assets, and the expertise, I want to impress upon you that Practical Law plus generative AI is transformative for three reasons. It's going to help us have our customers do even more with less. We can move from content to task execution. We can do that right where they're working, right within the workflow. Now, Practical Law has always provided trusted guidance to our customers to help them practice more effectively and efficiently. We do that with a very large team of experts that are creating and maintaining guidance notes, checklists, and model agreements.
This proprietary content set is infused with actionable expertise that goes beyond what the law is or how to understand the legal issues and tells our customers how to do the task at hand and even provides the model contract language to use in practice. Our competitive advantage is this unique, high-quality content set that is our moat and the experts that are behind it and the fact that our customers trust us that this content is up to date and accurate. Our customers tell us that Practical Law is like that trusted colleague down the hall that they go to for help. In-house counsel tells us that they save money on outside counsel using Practical Law. We've sold Practical Law for many years now talking about with the return on investment that you can do much more with less.
We talk about how these practice notes, these standard documents, this content set is like having that team in your office. AI allows us to amplify that in a number of ways. As you see here from the quotes on the page, our customers talk about being able to go well beyond the expertise they walked into the legal in-house department with or the small law firm attorney that can build that team of lawyers just from inside their computer. But with AI, we can help them find the answers even faster in a conversational interface using plain English questions. We can help them do that right within the workflow. So we can take the expertise out of the content, meet them in Word, in Teams, in CoCounsel. We can go the next step, the next mile, really, and help them do the actual task at hand.
So when you think about that message we've been saying for years of do more with less, we're now moving into legal task automation with Practical Law. I'm really excited about where we are today. But as Mike said, as David said, this is just the beginning. There's so much we can do here. And so I'm going to show you what we launched in January. And then I'll give you a preview of what's to come next. But let's first make it real with a scenario. So let's imagine that we're a senior counsel. And we've been asked by our client to draft an employment agreement for an incoming CEO based in New York. And that incoming CEO, they really want to ensure that the non-compete is enforceable.
So actually, what I need to find out now from Practical Law is what I need to know about how to draft that employment agreement, what's specific about New York, and the facts of what my client asked me, focus on that non-compete. So let's go ahead and take a look at the video and see how that will work in our new product. So starting at the welcome screen to Ask Practical Law AI, I can simply ask the question that I want to know. I don't need to think about keywords or what the right search is. I'm drafting an employment agreement for a CEO based in New York. How do I ensure that that non-compete is enforceable?
What the AI is going to do using that RAG methodology is search through all of the Practical Law resources that are relevant, digest them, and synthesize them into this very CLEAR answer that tells me, "In New York, here are the different facts you need to think about." I can then see very CLEARly, as this is generated only from that trusted Practical Law content, looking at the footnotes directly where it's coming from in the resources. If I want to take a deeper dive and continue on my Practical Law workflow, actually go into this checklist and dig a bit deeper and see, "Well, how long can this non-compete last for to be reasonable in New York?" Learn here that six months has been upheld by the courts. But now I want to take a step back and see what else might I need to think about.
So I can go to the Knowledge Map in Practical Law and see all of the related issues and content to this task at hand. I can look into a practice note on drafting executive employment agreements generally. But given that I've had that answer from GenAI, I'm feeling pretty good. So I'm going to go ahead and pull up that model agreement, the employment agreement, and see if this is a good starting point for me, drafted by Practical Law experts. Now, I know from my research that this is a good starting point, but I need something specific here to deal with that New York issue and that non-compete. So I'm going to go ahead and open up in Word and start tailoring my agreement. Now, in the past, this is where Practical Law would leave you off.
But now, with our new tool Clause Finder, I can find highly relevant clauses that meet my needs. Looking at all of these provisions and data points that our experts have trained AI models on, I can go in and stick with my example of a non-compete. So I'll go into my restrictive covenants. And now what I can do is find highly relevant non-compete agreement clauses for my particular example. I can search across three databases: Practical Law, publicly filed agreements with the SEC, and my internal collection. Starting with the gold standard with Practical Law, I can get to the drafting right there in my Word experience without even having to leave the page. But in my case, I know I want that 6-month duration.
So I'm going to go back and actually take a look and see what I can find in the SEC publicly filed agreements right from Word without needing a separate tool to search through those filings. I can limit this by six months because that's what I know I want to think about for New York. Go ahead and pull up those clauses with that six-month duration, take a look at them really quickly here in Word. And then I can go ahead, insert it into my draft, and get on with my work. So hopefully, as you see, by putting together Practical Law and AI, we're building on that value proposition. We're going from our expert content to doing the work. We're finding the answer faster, executing the tasks within the workflow. But as I said, we're just getting started. So what's next?
In 2024, we will continue to invest in that US experience, building on the technical capabilities, deepening the skills, and continuing to invest in the overall experience. We're going to bring Ask Practical Law AI to our other geographies where we have local Practical Law experiences: the UK, Canada, and Australia. We're going to integrate. We're going to integrate where the customer is working in Word, in Teams, in CoCounsel. This is so important when you think about the adoption of legal technology. Our customers can find these trusted answers, can execute these tasks without having to think about a tool or a feature or even know what products they have.
They get right within their workflow, accelerating that adoption curve. Let me leave you with this. Practical Law plus generative AI is transformative for three big reasons: building on that trusted expertise and content. We can help our customers do even more with less. We can move to executing tasks from content right within the workflow using a product that they've trusted over the years as their colleague down the hall. So with that, I'm going to turn this over to my colleague Rawia Ashraf, who's going to take a deeper dive into drafting.
Good morning, everyone. I'm Rawia Ashraf. I am the Vice President of Legal Tech Product Management here at TR. I'm also leading the Casetext and CoCounsel integration into TR. Today, I'm going to talk about what we're doing around drafting. This is a solution that we're going to bring to market later this summer. I really want you to think about two things as we talk through this. Drafting is a frequent and complex part of a lawyer's day. It's currently a not-well-solved problem. Second, TR, because of our unique content expertise and capabilities with generative AI, is uniquely positioned to solve this problem for our customers. I'm going to walk you through how we are going to do that. In the big, our clients are always asking us to meet them where they are.
So we have trusted products like Westlaw, Practical Law, CLEAR. But a lot of our attorneys and our clients spend time in other solutions and other suites of tools, including Microsoft. So a key part of our legal tech product strategy is to integrate with those surfaces as well. Here, you'll see we're integrating with Outlook, Word, and Teams, bringing CoCounsel and our content capabilities to those services. I'm going to dig in a little deeper with that Word integration. So as I mentioned, legal drafting is a complex part of a lawyer's day. Our studies show that lawyers spend between 40%-60% of their day, in fact, drafting documents in Microsoft Word. And then 96% of them feel like their drafting tools are not efficient. A full 96% are frustrated by this task that they have to perform for 50% of their careers.
Based on that, 82% of them fully anticipate investing in generative AI to help solve that problem in the next two years. We are gearing up to solve that problem for them. Among the things that make drafting so complex is that you have to start with a precedent or an example document. Lawyers rarely want to start with a blank page. They want to build off the knowledge of their firm or their client's previous engagements. Second, once they find that starting point, which can often take hours, they have to manage a lot of different inputs. They're bringing in research from Westlaw, content from Practical Law. They're getting comments from the other side or from the partner they're working with.
Then they have to follow either court regulations on how a document can be filed, or they have to make sure all of the agreement is internally consistent with a ton of different authors in the mix. So we are tackling all of that for them with our Intelligent Drafting solution, which aims to reimagine the drafting experience. It's going to be an end-to-end solution that gets you from the starting point all the way to finalizing a draft. So there's a few things I want to highlight about this and what really sets us apart and what's differentiated in our solution. First, it's going to be an end-to-end solution. It's not piecemeal. It doesn't pick off a couple of parts of the flow. It's there from the start to the finish of the legal professional's experience.
Second, it's going to serve multiple use cases, so both transactional attorneys and litigation attorneys and, in fact, a bit of advisory work as well. What's really unique about our solution compared to everything else you'll see in the market is that it's grounded in TR content, so the content that Mike and Emily walked you through and the RAG capabilities that we can bring that we can support through our trusted content. And finally, what's really exciting is that it's going to be powered by the CoCounsel AI assistant. So hopefully, what I'm going to show you in this demo starts to look familiar. You should have seen it in what David was showing, Emily was showing, and Mike was showing because we really want this consistent product experience. Let's take a look at how this works. So I have the CoCounsel AI assistant open in Word.
And I type in, "I need a sale of goods agreement." Now, as I mentioned, lawyers rarely start with a blank page. So what we get to first is a document automation capability where they can enter some data points from a template that the firm has built through Contract Express, which is our document automation tool. You can use Practical Law examples here. You can use the firm's own internal examples here as well to get started. So we'll fill in that data. And then we'll get a first draft of a sale of goods agreement. Now, this is just based on a template. And I know there's deal points in here that I have to change based on the negotiation or the relationship with the client.
So here, I'm asking CoCounsel to find examples of indemnification clauses because I need to change that or add that clause into this agreement. What's happening here is that we're triggering Practical Law right in the drafting experience, as Emily showed you. I'm getting examples from Practical Law content. I can get examples from SEC agreements. Really importantly, I can get examples from my own content. Here, we're allowing the users to bring their knowledge and Practical Law knowledge into one place so that they can quickly get on with drafting and negotiating. Let's say I take that and I put that in if I'm happy with it. Now, I know I need to modify the reps and warranty provision. I want it to cover intellectual property rights. I'm asking CoCounsel to modify the existing clause in the agreement. It's context aware.
It knows what I'm talking about. It creates a modified draft. Now, what we're doing in this and we're not just hitting GPT-4 and pulling back what ChatGPT would say about modifying a reps and warranty provision. We are taking Practical Law drafting notes and sample Practical Law language. We're bumping that up against the user's query here. And we're sending that to the large language model. And we're saying, in the same way that you would tell a junior associate to look at Practical Law before you draft something, we're now telling the large language model to use Practical Law guidance when you're responding or when you're generating. Here, you will also pinpoint the sources that we've shared with the large language model so that you can double-check the work yourself. So let's say we like this one. I'm going to keep going.
I'm going to add it in. Now, I also need to do some legal research on the fly. I want to understand if this limitation of liability is enforceable, similar to some of the research that Mike was showing, hitting Westlaw now from the Word experience. And you'll see here, I'm getting a Westlaw synthesis of how the law applies here as well as citations so I can click into Westlaw and do additional research if I want to do that to verify what the AI is telling me. Let's keep going. Let's imagine in this scenario, it wasn't me that started the draft, right? I received a contract from a third party. So I'm now representing my client and negotiating it.
So for my client, I have a playbook of preferred language or policies and requirements they need to see in every sale of goods agreement before they're willing to sign it. So I'm going to compare the draft I was sent by the third party to my playbook. I represent a buyer. Here, the intelligent drafting tool leveraging that playbook is telling me, "Hey, this contract is missing a set-off clause." And a set-off clause is something this client requires in all of their sale of goods agreements. And what this is doing is solving a problem that's really difficult for lawyers. Lawyers can look at clauses in agreements and understand if there's something off about it. But it's very hard to identify what clauses are missing, what are the things that you weren't thinking about that are essential. So here, we're bringing that in.
Then we're giving them the preferred language that the client wants to see in all of their sale of goods agreements. We're also giving them fallback language in case they have to negotiate that deal point a little bit more. Now, finally, let's say we've done all of the negotiations. We've gotten comments back from the other side. We're ready to finalize this agreement and send it for execution. What's critical here is that there are a lot of legal errors that would be detrimental to a lawyer's credibility. I'm not talking about spelling mistakes or proofreading issues, but legally substantive mistakes, right? So what we do through our tool is allow you to check for those as well. We'll run that Deal Proof scenario.
We'll come back with all of these legally insufficient issues with the contract that the lawyer then can fix really quickly. So example, the terms are not cross-referencing properly, or the definitions are contradictory, things that would actually hamper the interpretation of the contract. So you can run through those and quickly fix them. We'll give you suggested changes. Or the lawyer can create their own fixes on the fly. Now, that's a look at the end-to-end solution. Parts of that are in market right now. The full solution will be launched this summer. It goes through all of these steps in a lawyer's drafting journey. It also connects to HighQ. As I showed you, it's rooted in the CoCounsel legal AI assistant. It connects to customers' own documents. Not only is it surfaced in Word, it'll also integrate with Microsoft Copilot.
So I just want to leave you with a couple of thoughts about drafting. First, drafting is a big part of lawyers' days. It's messy. It's complex. It's not a well-solved problem right now. Two, we feel like we are uniquely positioned through the solutions I've shown you plus our content and our generative AI capabilities to really solve that problem for our clients. With that, I'm going to invite Melissa Oakes up to come talk about our tax research product, Checkpoint.
Hi, everyone. I'm Melissa Oakes. I am the leader within our tax product team. I'm excited to talk with you today about Checkpoint Edge and generative AI. Now, we've heard a lot today about legal research. At a high level, tax research is similar. To advise your clients, you need to understand the relevant laws and regulations and how those will be interpreted by the IRS and the courts. But there are a few ways in which tax is unique. First, many tax professionals are accountants, not lawyers. They didn't go to law school and often don't have any formal education in interpreting laws or applying case law to their client's situation. At the same time, our customers in accounting firms are increasingly moving into more complex advisory services that require a more sophisticated and strategic understanding of the law.
On top of that, we've seen an unprecedented rate of change in tax policy over the past six or seven years, which means it's hard for even the most seasoned professionals to feel confident that their knowledge is up to date on any given day. So all of this informs what tax professionals need from their research solutions and how they want to interact with their content and tools. So Checkpoint Edge, let me give you a little bit of background about us. Checkpoint Edge is the leading tax research and advisory platform. Our content is trusted by more than 200,000 tax and accounting professionals. We, of course, have all of those tax laws, regulations, and cases that form the basis of tax law.
This is all robustly interlinked with our proprietary content, which is created and maintained by subject matter experts who have deep domain expertise and practical experience. Resources like plain language explanations, practical guidance, and workflow tools are essential for tax professionals being able to understand this really complex space. I was a tax lawyer before I joined TR. And I use Checkpoint every single day as an essential tool in my job. The hardest part about tax research today is research initiation, getting started, building your foundational knowledge of the issue you're working on, and ensuring that your knowledge is still up to date. Those rules are changing constantly. We hear from firm leaders that junior staff, in particular, default to starting in Google because it's just where they're more familiar. And so I'm going to walk you through the demo.
You'll see how Checkpoint Edge and generative AI is going to be transformative for our customers. First thing you'll see is that there's no need to do any upfront decision-making. Put in my question. I'm going to get back an answer that's easy to read and that's grounded in trusted, accurate content. If I'm new to this topic, I now have that foundational answer that's going to set me up for success as I move forward. If I'm experienced in this area, I've now confirmed that my knowledge is still up to date and accurate. I can see the citations to the trusted content that I'm familiar with. I can click through to those citations and validate the answer, find more in-depth explanations, practical examples, sample documents. I can see the key section of the tax code that applies to my issue.
This is really important for tax professionals. There's a mantra in tax that you should start in the code. That's really hard to do if you don't know what code section applies to your issue. I can click through to related resources. This is going to help our customers uncover the content and tools in their subscription that they may not even know that are available, but they've been purpose-built for the task they're working on. We're making it easy for tax professionals to get where they need to go. We're automating and integrating that entire pipeline from research to workflow. Customers will be able to click through to our topic pages, another feature we're launching this year.
This is going to meet customers where they're at in their workflow, whether they're just learning about the topic for the first time, working on a complex advisory engagement, or filing a return. We're connecting them with the resources that we've built for them to be able to get the job done. As additional questions come up in my research, I can pop back to that AI conversation, ask those follow-up questions. And again, I get an answer that is grounded in trusted content. It's easy to read. I have the step-by-step actions I need to take to serve my client. We had our alpha testing late last year. We also had our alpha prototype available at our user conference in the fall. And I can tell you, our customers are ready for this.
Over half of our alpha testers said that AI-assisted research would become the primary way they start tax research in the future. We had folks wanting to buy it on the spot as-is. So easy to use and intuitive. I can just dive right in. But most importantly, I know that I can trust it because it's grounded in Checkpoint's comprehensive, accurate content. It's important to remember that a tool like ChatGPT has static knowledge. I believe the most recent update was around April 2023. As a tax professional, I can't be relying on information that is, at best, a year out of date. Our Checkpoint content is continuously maintained by subject matter experts. Our coverage of new tax legislation consistently beats our competitors when it comes to the speed of coverage and the depth of analysis.
Because we have superior content, we have the workflow tools that help the customers get the job done, we're creating this end-to-end experience that is going to take the complexity out of tax. We'll be launching AI-assisted research in Checkpoint Edge for federal tax this summer. But that's just the beginning. We'll be adding on our state jurisdictions and continuing to automate that full pipeline. And we'll be building on our AI platform and AI skills that we'll have at TR. And with that, I will turn it over to my colleague, Piratta, to tell you more about SurePrep.
Hi. I'm Piratta Van Rijn. I'm our Global Product Leader for our tax and accounting products. I've been a digital product leader for over 15 years in fintech and SaaS in this industry, the last three here at Thomson Reuters. It's not often we get to say it's exciting times in the tax and accounting industry. But it is really exciting times in the tax and accounting industry. Let me tell and show you why. Just as a reminder, our customers are accounting firms that range from some of the world's largest to the neighborhood accountant in the local office. Their biggest pain point is a shrinking talent pool, faster and more expansive legislation, more demands from their clients, deadlines that never move. So they are literally trying to do more with less each year. Their number one need is efficiency.
That's where Thomson Reuters and SurePrep come in. Today, I want to showcase the SurePrep suite of products that we acquired in January 2023 to automate the process of preparing taxes with the power of AI, machine learning, and other automation technologies. SurePrep is a purpose-built suite of innovative products that saves accountants time across their workflow. We eliminate countless manual steps. We save accountants hours on complex returns. While SurePrep itself is not tax software and in Thomson Reuters, we have UltraTax, GoSystem Tax, ONESOURCE Income Tax for that, the power of SurePrep is how it seamlessly integrates into those tax products, including those outside of Thomson Reuters. Before I move into the product demos, just a really quick high level about what is this end-to-end workflow.
So it includes three steps: first, gathering tax data and documents, then preparing and reviewing the returns, and eventually filing and delivering those returns. OK. So let's move to the demo. So in this demo, I'm going to be playing two roles. You will see me as an accountant but also as their client, Tom, taxpayer. And this might be a role that you guys can relate to. I'll start at the beginning of the tax workflow, which is gathering data. This is actually one of the most painful parts of the process because it depends on the client gathering the documents and uploading them, providing information about changes that could impact their taxes. And so here I am as the client, Tom, taxpayer, generally not motivated to start my taxes until I get a reminder from my trusted accountant.
I'm really glad to see that TaxCaddy makes it super simple and secure to log in because my accountant has already verified me. This is my dashboard. I've got a few files here on my desktop where I scanned in my tax documents. It's easy to drag and drop. But what's really interesting here is I can see TaxCaddy is able to recognize what kind of tax document this is, break apart a PDF where I had a bunch of files, and check them off the list. So now, I'm going to step out of the client role. As the accountant, I'll get a notification. Now, this is good news for me. I see my client is getting data and documents faster to me than ever before.
That means I can start, and my team can start preparing those taxes faster and before we run up to the deadlines. This is my dashboard in TaxCaddy. I can see those documents that my client, Tom, taxpayer, has just uploaded. I'll be able to see also what's still missing. But what I'm really excited about is the green auto button because that means TaxCaddy has done the work of already identifying the documents and where they belong. Now, this can save 15-45 minutes per return. Can you imagine clients with hundreds of documents coming in that would have to be manually moved to different places in the return? I'll take a look and start flagging the documents that I want to move over into the tax return and filter out those things that clients send me that I don't actually need.
Click Create Binder. And then I'll go ahead and be able to move to starting to actually prepare and review the tax return. So let's go on to 1040SCAN. So here we are in 1040SCAN that uses our patented technology that allows us to automatically capture and verify data from more tax documents, from more fields than anyone else out there. So this is much more than just OCR. That's just our base technology. On top of that, we have built in layers and layers of business logic for tax treatments and connections to tax software. So let's take a look at how it works. Once we open the binder with those documents that just came through and from TaxCaddy, we can see a K-1 on here.
So what we have is we have the extracted data over on the left-hand side along with the source document on the right-hand side. It's important still to do that verification side by side. The colors are going to guide the user's focus. The pink means we auto-verified it. SurePrep already auto-verifies 68% of standard documents. We keep growing that. Yellow, it's not yet auto-verified. This needs to have a look. This can be done by the accounting firm's team. Or they can outsource to our expert-trained team at 1040SCANv erify. So on this K-1, you can see the address is highlighted in yellow. We're going to make an update before we consider this reviewed, hit Submit, and move forward to SPbinder or SurePrep Binder. SurePrep Binder is our custom work paper tool.
This is where accountants do the work of annotating and ticking and tying information before it goes to the tax return. Here in SPbinder, also at the bottom of the screen, you can see SurePrep's API is now proposing tax treatments for line items in this complex hedge fund K-1. Now, let's look at another example of a non-OCR document coupled with our live sync functionality. On the left-hand side, we still see SPbinder. On the right-hand side is our GoSystem Tax Tax Engine. This is where the data eventually needs to end up. If it's done manually, it's slow and error-prone. So we created live sync that moves the data automatically between the two systems. Here we'll be able to see that live sync is active and ready to go.
In the SPbinder Forms tab, we still need to first capture the client's business income from the income statement. We do that by highlighting it. When that happens, we're actually OCR-ing it in real time so users don't have to type it in. Once that field is saved in SPbinder, live sync starts to sync into the tax software in real time. You see that same amount appearing in the tax return on the right-hand side, again, saving time and steps from bulk updates. The great news is that once the data is captured in SPbinder through those OCR highlights, our system is learning. So that means next year, we'll be able to recognize this document type and be able to auto-verify it. That's another key differentiator where SurePrep is saving accountants valuable time and effort, up to 45 minutes on complex returns.
So now, the final step, deliver and file, which is also surprisingly time-consuming and full of unexpected delays. As the accountant, I've logged back into my dashboard. I can see the returns that were ready to file. If we go back to the last thank you. I'll choose Tom, taxpayer, and prepare his documents. First, I fetch them from the tax software, so pulling in the tax return and authorizations, payment information, scan it over before I approve it all, and just hit Deliver. That's it. Now, I'm going to step back over to Tom, taxpayer, where he would have gotten a notice on his mobile phone. As Tom, he would have been able to skim through that document and see the return and sign it off.
Once he gives permission, we are automatically, behind the scenes, transmitting it to the IRS and the state revenue agencies. Essentially, we have automated the e-file and delivery process. Now, both the accountant and Tom will be notified when those returns are accepted. That's another 20-30 minutes off the work that the accountant had to do on returns. This is just the beginning. We have so much more work ahead of us to automate and get rid of even more manual steps, using generative AI to continue to make that work easier and faster for accountants so they can be more efficient, especially when it comes to the work of tax prep, so they can start to work on more higher-value tasks.
To summarize, SurePrep is a game changer for efficiency. That is our accountant customers' number one need. We've already proven that we can save accountants valuable time in minutes and hours in preparing returns. We're moving at pace to deliver even more time savings in the coming quarters. With that, I'll wrap it up and welcome Jake Heller, Head of Product for CoCounsel, to wrap up our demos.
Before I get started, amazing stuff so far, right? I hope you're as excited as I am. I'm Jake. I was, until recently, the co-founder and CEO of a company called Casetext, a company I led for about 10 years. I now proudly joined Thomson Reuters through the Casetext acquisition, where I am leading a product for something we're calling CoCounsel. Today's presentation is really going to go over two things. The first is where CoCounsel is today, how it is today supporting lawyers in everyday tasks. Secondly, where it's going and the impact we hope to make, not just for lawyers but across all of the professions that Thomson Reuters serves. So what is CoCounsel? Today, CoCounsel is what we're calling an AI legal assistant.
At Casetext, for about 10 years, we were working on the application of artificial intelligence to law. For the last couple of years, we started this concept of an AI legal assistant, something we're continuing on with our work here. Really, an AI legal assistant, we think, means two things. The first is that you can work with it like you work with a colleague. All you have to do is talk to it like you would another person at your law firm or anyone else in your office. And it understands what you want it to do and gets to work. Second and more importantly, is that you can delegate to it substantive legal work and expect it to get that work done, not just at a human level of quality but because we're using the power of machines, right, at superhuman speed.
I want to give you all a live demonstration of this so you can see what it looks like to work with this in practice. So let's switch over to the computer. Here we are. This is CoCounsel. It may, on the surface, look a lot like other chat and AI tools you've seen in the past. But I can promise you, it is nothing like it. For the purpose of this example, I'm going to pretend that I am a lawyer representing Microsoft and their recent successful acquisition of Activision Blizzard, a big gaming company. And what I'm going to attempt to do here is give it about a week's worth of work in about 5 minutes and watch what it does, how it's able to complete that work in about 15 minutes.
So I'm going to start off just by talking to it, telling it, "Hey, I'm a lawyer from Microsoft dealing with this acquisition." I'm going to ask it, "What are the different ways it can help me?" And it's listing out all the different things it's able to do and asking me, "Hey, well, here's what I can do. What would you like me to do for you?" So I'm going to start by saying, "Let's start by reviewing the merger agreement." It's saying, "OK, fine. No worries. Upload the merger agreement." I'm going to start by working with both the merger agreement and a letter agreement that amended the original merger agreement. And we're going to ask some specific questions that I've already prepared, things like, how much is the parent transaction fee? Under what specific circumstances can Microsoft be required to pay the reverse termination fee?
And things like that. Of course, no AI demo is complete without something going wrong with the Wi-Fi. So I'll try to fix that. Well, this is why we have videos, just in case. Oh boy. So maybe we should switch over to the video. Sorry about that. Oh, well, no, we're back. I'm back on the computer. I'm going to switch back to that. Sorry about that. So I'm going to start by asking it the things that specifically I'd like to know and give it the documents that I'm working with. Spoke too soon. Back to the video. So what the AI is able to do, and I will walk through it with the video in a second, is it's able to understand what I'm asking it to do.
As you saw already in the initial part where it did work, it was able to suggest that I upload the documents, suggest the work it's able to do. And if the team was able to switch back to the video, thank you. And I'm able to kick off a pretty intense review of the documents. I'm also able to say, "Hey, I'd like you to review what is review the evidence from the case." Or in this matter, it could be for due diligencing, whether or not we ought to take it. So I'm asking questions like, is there any evidence that Microsoft is pursuing its decision here, its acquisition for anti-competitive purposes? And it's reviewing literally 100 documents to get that work done. I'm able to ask it to kick off a review of those documents. And here, it's pulling up that information.
What would have happened if the Wi-Fi had worked is what you've seen me do is give it, for example, these very difficult questions on the merger agreement. For it to come back saying the termination fee is $3.5 billion but goes up to $4.5 billion on certain dates, it would have created a timeline of all of the different events in the merger agreement, spelling out step by step by step what happens when and what are the obligations. It would have kicked off a review of the documents and walked through all the different in this case, for example, the different international rulings in different languages, which I'm able to ask English questions against. Here, it is answering how, for example, the Japanese Antitrust Commission has made its decision.
And you can even verify that just in case you don't trust the AI, it can verify here that it's getting the exact right translation of a Japanese document and answering difficult questions, like what its conclusion is. And you're also able to, again, review literally over the course of minutes, hundreds of documents that go into detail, internal communications within Microsoft that were a part of our due diligence in this case, where in a chat transcript, it turns out they made some comments that would definitely be worthy of deeper investigation, for example, whether or not they're going to force exclusivity of some of their games and whether or not they should, therefore, take PlayStation out of their economic model of competition, right, so something that would have raised flags for an investigator trying to make sure assess the risk of the deal, for example.
And finally, it would have summarized point by point every single one of the pieces of the briefs filed in front of the FTC. And again, this is all work that you can kick off over the course of just a few minutes using advanced AI like CoCounsel. So given that power, we are excited to share that we are now the leading application of artificial intelligence to law. We have, over the last year alone, CoCounsel has read and processed 640 billion words. And it's completed over 1.1 million tasks for lawyers. This is an example of generative AI in use in practice right now by professionals, tens of thousands of professionals. By the way, we did the math.
If you printed out all of the pages that CoCounsel has read, done work for lawyers, it would wrap around the world not just one time but nine times, right? That's how much it was able to read over the last year. What makes it special is a few things. First of all, it's been designed from the ground up as an AI assistant built specifically for law. It has broad skills and capabilities. As you saw in the, unfortunately, demonstration video, is that it is capable of reading difficult contracts, assessing risk by doing due diligence. It was able to do research against international documents and read in multiple languages to answer difficult and substantive questions. And we're just getting started. It is grounded in real legal documents, something you've heard almost every presenter here talk about so far. It's something that's very important.
It is going to be drawing its answers not just based on its memory but based instead on the actual documents that it's reading, mitigating the risks of inaccuracies. And we'll know AI is perfect, like no human is perfect. We have thoroughly tested it with over 600,000 tests, making it accurate, verifiable, and trusted, verifiable because every single thing that it says will be cited to a source that the person can review and assess for themselves. It's also operating at an unprecedented scale. What we did over the course of the demonstration is had it read over 10,000 pages of documents, including a 500-page Japanese document on their kind of Fair Trade Commission ruling, a 400-page document from the United Kingdom that actually came out the other direction at first, hundreds of pages of briefing, and thousands of pages of discovery and internal risk mitigation through compliance checks.
And then finally, we have done a very serious focus on security and privacy, something that matters deeply for legal professionals when we're working with their most sensitive documents. But we're just getting started. I want to talk a little bit about what's next. As you heard David present on, this technology, the ability to work collaboratively with an AI assistant, is going to become TR's AI assistant for professionals, not just legal professionals but professionals' heartbeat. What you'd expect to see from us is, over the near term, we're going to be working to integrate and work alongside, as you saw in some of the demonstrations, not just the legal products but across the board of different TR products that we're working together with CoCounsel.
It's going to be accessible in more surfaces, not just the web interface that you saw today but also in places like Word and Teams and so on. We're going to be expanding to new geographies. And I'm excited to announce that today, we have launched into the United Kingdom. We're going to launch into Canada. And we're going to Australia next. This AI is already quite capable. What we're calling skills, right, its capabilities across the board today includes reviewing documents, reviewing contracts, doing research, et cetera. But in the very near future, you should expect to see it do far more for our legal professionals.
And all of this is going to happen just this year in 2024. So to wrap up, the AI assistant today is in a really, really exciting place. It is already capable and has already done over 1 million tasks for lawyers across the country and is already the most skilled, most secure, and capable AI assistant. But just getting increasingly good and is going to be something that we hope impacts every professional who uses TR services. So now, I'm excited to reintroduce and reinvite Steve, Mike, Kirsty, David, and Gary back to the stage for our first Q&A. Thank you.
All right. I think we have some mics out in the audience somewhere. It's hard with the lights to see. I guess we have Mike back there.
I think Heather was first.
Heather?
Yeah. And then Manav. And then Andrew.
Hi, Heather Balsky, Bank of America. Thank you for the question. I was curious, actually, about the drafting tool that was presented. My understanding, and please correct me, is that you don't have a material presence in drafting today. So I'm curious how much this adds to your TAM, how you're thinking about the opportunity long term. And also, it seems like the tool prompted a lot of your existing legal tools. So I'm curious, are you going to market with this as a bundle with all of your legal tools today? Is it a separate tool? Just how all that works. Thanks.
David, do you want to start? And then I'll add.
Well, let me start by just setting the foundation of where we are today. So we do have existing tools for drafting, starting with content, such as Practical Law's standard documents, Clause Finder, and a few of those content sets, which are used at the very beginning of drafting. And we have a few tools which are already available, integrated into Word, things like Drafting Assistant and some of the document automation tools which Rawia highlighted. What we're really doing with this launch is we're tying it together. So as Rawia spoke about, we now have solutions for the end-to-end workflow of drafting. And that's what we're going to be launching this summer as the Intelligent Drafting Solution. And so it's a repackaging and an expansion of the solutions that we're going to have for drafting in the marketplace.
I think the only thing I'd add, Heather, is I think when you look at what Mike Dahn walked through with Westlaw Precision, I mean, one of the big punchlines from that, which Emily is already onto and has been for a while with Practical Law, is moving out of research into research-driven work products. And drafting, I think, really brings that to life. And so what you're going to see us do is sort of two things. One, extend our leadership in research into producing the first draft of work products for lawyers and for tax and accounting professionals and ultimately for risk professionals.
And then the second is bring these things together so that they're increasingly indistinguishable. And we think that's a pretty big opportunity. And what we've quantified today is very much, as I said, existing products, product roadmaps, and existing prices. But over time, we've got our eyes on being able to automate and perform more and more tasks to free lawyers up for the value-added stuff that, as one said, they went to law school to do.
Rawia, anything you would like to supplement?
Is this on? Yeah, I think David and Steve covered it pretty well. I think what's net new, what we're bringing to market, are all the generative AI capabilities, the playbook comparison, those things that add on top of what Practical Law and Westlaw can do and what we could do with proofreading. Now, we're bringing AI into drafting as well.
I just want to clarify. You did prompt a fair number of your existing tools. I wrote down every tool you were using. So can you use the drafting tool and, say, not have Practical Law? Or is it one of these things where you want all the tools, and then you can use the drafting product?
It's optimized if you have all the underlying tools. There's certain parts of it that just rely on a firm's own content that don't require our content. Those you can use. But it's really the end-to-end solution only works if you're investing in Practical Law and Westlaw and our other solutions.
That's helpful. Thank you.
I think Manav was next. Just here, Julia. Thank you. And then Andrew afterwards.
Thank you. So a lot of cool stuff that you guys demoed here, with GenAI and stuff. The question is, how are you pricing all these things to your customers? And how much of it is going to end up being table stakes where maybe it's just a better renewal tool as opposed to actually charging? And then just as a follow-up, Steve, you mentioned the incremental $12 billion was current pricing. So just as a follow-up, can you talk about how to think about what that current pricing is and the trend in a couple of years, I guess?
Well, number one is that we have already captured, I think, meaningful uplift in revenue from some of these GenAI tools as part of Westlaw, for example. So Westlaw Precision includes the GenAI features. And we've been able to get increasing premiums on Westlaw Precision as a result of rolling in these features. So we strongly believe that, in our leadership position right now with generative AI, that we're going to be able to garner premiums from our customers, that it won't be just table stakes.
For some of the other features, we're going to take these opportunities kind of one at a time. And we're going to essentially establish what we think is the value we deliver to our customers and price accordingly. As Mike has mentioned, we are taking a very disciplined approach here to try to make sure that we are reflecting within our pricing the value that we deliver to our customers. So I don't know, Steve, if you want to add anything?
Yeah. I mean, the only thing I'd add, Manav, is what we've tried to quantify this morning is that current roadmap, which results in some price uplift, which results in some, in a sense, new customers coming online or at least new use cases. What we haven't priced in is sort of the displacement of labor and so very significant sort of shifts away from headcount into technology spend. Because, as I said in my remarks, it's really too early to do that, both in terms of what does that end state look like and what will the sort of transition path to that, how long will it be? Andrew.
Hi. It's Andrew Steinerman at JP Morgan. David, I know we're using the word CoCounsel, Jake's company, to be all the legal assistants at Thomson. But when I look at the products, including Westlaw Precision, I believe, it feels, that there's one legal assistant at Westlaw Precision and a separate module still at Casetext. I know you only acquired it last year. Is that accurate? And when will there be only one legal assistant in all of Westlaw? And will that convergence cause any self-capitalization of revenues?
Yeah. one, thank you for paying very close attention to the demos because you're absolutely right. Yes, today, what we wanted to do was we shared, at the very start, a vision of how we want this all to come together. So the demo that I walked through was a completely integrated CoCounsel experience with the TR products. Where we are today is, you're right, we have a skill which is inside Westlaw, which is called AI-assisted research. And we have CoCounsel, which is a separate interface. Now, we are running furiously right now on the integration to get towards a place where we have one consistent experience and AI assistant which is running in the background, whether you're using Westlaw, Practical Law, or CoCounsel. And we anticipate to deliver that this year in 2024. So that's number one.
In terms of cannibalization, we don't believe that there is cannibalization here at all because the skills are distinct between the different products. So each of our products brings their own unique capability. Practical Law brings know-how. Drafting brings a drafting solution. Checkpoint brings tax. Westlaw brings case research. And CoCounsel is developing independent skills like email drafting, document analysis, and future skills which Jake's team are working on. And these are going to be independent. And what we think will be the case is that, as you add more to the assistant, that it will become more valuable, potentially, again, as that labor displacement. Because if you have an assistant that does one thing, it's very valuable. If it does 20 things, all within a particular professional domain, it is more valuable as a whole. That's our hypothesis.
The other thing I'd add, Andrew, is Elizabeth will mention this in her presentation on tax and accounting professionals. Laura will mention it as it pertains to corporates. But Jake, as he said, has started in the legal profession. But we're pretty excited about taking both the CoCounsel brand and the CoCounsel skills into the tax and accounting space, whether that's for sort of tax and accounting firms or whether that's for the head of tax. And so you'll hear us talk about that as we go through those presentations as well.
Tony up front, Beth, at the first table. Tony Kaplan.
Hi. Tony Kaplan from Morgan Stanley. I was hoping you could share any sort of usage stats that you've noticed with regard to when people have actually adopted the AI tools. Are you finding that they continuously use them? Are you finding that maybe they start off using them, and then it drops off? And so just trying to understand any sort of usage behavior that you've noticed that's changed.
David, do you want to start that, or Kirsty?
Kirsty, you want to start?
So I think, Tony, they're quite consistent so far from what we've seen, in line with, as you'd expect on a kind of working day, right? So the middle of the week seems to be the busiest. But we see, particularly in Westlaw, we see people coming back again and again and again. I'm not sure it always amuses David. I spend a lot of time looking at that because I have to manage the overall cloud costs. So it's something we literally are looking at every day. But so far, it's been really delightful to see how much it's picked up and how consistent it is.
Yeah. Maybe to add, well, just to step back, we are taking, again, I think, a disciplined approach in evaluating the success of each of our different launches in that we're applying a product analytics and a growth accounting method, which both Jake's team, my teams, have had experience with from other companies and within TR. What we look at is first-time user adoption, retention, and whether we see evidence of product-market fit with every single feature that we launch. Those that work, we put energy behind. Those that don't work, we stop, and we move on. So that's one thing just to start with. The second is that, with TR, because we have an existing distribution footprint, we've been able to bootstrap a lot of our usage faster and more effectively than others in the market.
When we launched AI-assisted research in Westlaw, we rolled it into the Precision subscription tier, which meant that we had those thousands of customers and tens of thousands of users on day one when we launched. This is very different than others. When Jake launched CoCounsel, he had to scale it up from zero all the way up to those 1.1 million tasks. And so we've been able to bootstrap, thanks to our distribution. And generally, we're seeing, for the skills that we're launching and we're putting energy behind, that usage is increasing. And we're seeing good retention.
Mike or Jake may want to add to this, Tony. But the only thing I'd add is, what's been somewhat surprising to me is, for example, the sole litigator in the Midwest saying, "I have to sign up to these TR AI products because I was thinking about hiring a paralegal. I've had paralegals in the past. I don't particularly like having to manage people. A paralegal was going to cost me $90,000 a year. This is just fundamentally a cheaper price point. And it works just as well." So that's one example. The other example is literally the litigator who says, "All right. I'm in court tomorrow morning. Here's the question I've got to address.
Mike Dahn, type it into Westlaw Precision AI. I'll compare that answer to that, which my extensive research had produced." So we're actually, rather than just penetrating the very largest firms with the biggest budgets or the largest general counsel's offices, what we're actually seeing is, I think, a pretty even demand curve in the early going. It'll be interesting to see whether that continues. But that's been exciting to see.
I think we have time for one more before the break. I think, Beth, right behind you, I see a hand by the door.
In the bright lights .
Thank you. It's Kevin McVeigh from UBS. Very helpful. As you think about the kind of go-to-market strategy, does the pricing have more of a consumption component to it, so kind of more of a fixed-end variable? Because I'd imagine, as particularly the GenAI becomes more and more embedded, the usage probably scales disproportionately. So is there any way to think about subscription versus maybe a variable component as the model scales?
So I'll turn this over to David and Kirsty. And Mike might want to comment. We have debated this every which way to Sunday, as you can imagine, Kevin. So because what's new about this is we are pinging large language models. So there's a variable cost component to that. And we need to make sure that we do that in the best interests of not only our customers but our shareholders. What we've settled upon, and I'll ask David and Kirsty to comment on the specifics, is the simplest possible sort of packaging and proposition that we possi bly can because you've got a set of customers who are saying
"I know my profession's about to be transformed. I trust TR to drive that transformation in my best interests. But I've got to figure out how to use these products and what it all means." If you ask me to figure out sort of very sophisticated pricing models and so forth and entry points, it's overwhelming. So we've tried to sort of simplify it as much as possible. But David and Kirsty.
Yeah. What I would say is, right now, because we are in an early stage, we're trying to, again, keep it simple so that we can drive adoption and usage to establish the most valuable use cases with our customers. And so, whenever possible, we are trying to make it, again, simple and to drive usage whenever possible. There are a couple of exceptions, right, where our customers have come to us with bulk document analysis or a few things where it's just a huge data volume. And there, we have explored usage-based models. But we've generally seen, with the interactive experiences, like we've shown on screen, that usage-based pricing is not really well received because it creates a disincentive for consumption and adoption.
Yeah. I would just add, from a platform perspective, hopefully CLEAR from my presentation, right, we work with different LLMs, partly because they're good at different things. They also have very different price points. One of the things that we're focused on, from a platform perspective, is optimizing those, obviously, firstly, for the product experience, but secondly, also for the cost and thinking about what those can do. We've set up the platform very much to be plug and play. We can actually swap in different models very easily. It's something we've already got quite a lot of confidence that we're going to really be able to leverage because we are agnostic. We haven't just signed up with one particular provider.
Kirsty touched upon this. But when she talks about the platform, it is one GenAI platform across the whole company. So one of the things I mentioned was sort of the opportunity for the Reuters newsroom to tap into that. Certainly, it is something that, as we push more and more tax products onto that platform, accessing large language models, we've got a pretty good ability, Kirsty in particular and her team, to sort of do that resource load leveling and make sure that we don't see sort of spikes that we can't manage because we've got all our businesses and all our products running through the same thing. That helps in a big way.
I think we should stop for now. We'll have a second Q&A later. So, with that, we'll go to the break. We'll be back at 10:30 A.M.
Thanks.
Thank you. There we go. Thanks. Welcome back after a quick break. So if everyone could take their seats, please. We'll get the second half of the program underway here. I'll start off by introducing Raghu Ramanathan, who is the new President of our Legal Professionals business.
Hi. Good morning. Good morning. My name is Raghu Ramanathan. I am the President of the Legal Professionals segment at Thomson Reuters. I joined Thomson Reuters about a month ago. Previously, I used to work for SAP, running their platform division, otherwise known as the Business Technology Platform, which I'm really proud to have grown from a $270 million cloud business to a $2.2 billion cloud business over the span of five years. One of my main motivations in joining Thomson Reuters is the incredible opportunity that we have to shape the legal industry, especially with the advancement of AI and my belief that this industry will be a signpost for other industries to follow. After my first month here, I want to say that I'm starting to feel at home.
I'm quite impressed with the talent, the passion, and the commitment to customers that I see at TR, in addition to the very exciting product roadmap. Let me start by sharing my three key takeaways for the Legal Professional business. First, we have a very exciting market opportunity before us. With an increasingly complex legal and regulatory environment and the rise of generative AI, the legal industry is primed to use more technology to help legal professionals do their work more efficiently and effectively. Secondly, Thomson Reuters is exceptionally well-positioned to capture this opportunity. Not only do we have a market leadership position, but we also have an early lead in launching GenAI products, as you've seen. And we're already demonstrating how AI can be monetized. And finally, driven by the first two points, the growth potential of the business is accelerating.
We see significant growth of our TAM. We are targeting 7%-8% organic growth by 2026. Now, let me go into greater details behind these takeaways. Let's start with the market. Today, with our portfolio of content and technology solutions, we compete in a $10 billion vendor market. We estimate the total addressable market for legal professionals to be at $23 billion. We see our market growing by 6%-8% annually. This growth is driven both by increasing penetration and adoption of currently available solutions and opportunities also created by new AI-powered solutions. One important characteristic of this market is that it's historically stable and recession-resistant. For example, last year, while normal business declined due to toughening economic conditions, countercyclical business, such as bankruptcies, grew faster to compensate. If you now look at Thomson Reuters, TR holds the strongest position in this market segment.
Not only do we have the number one position by market share, but we also command a sizable price premium and have the highest customer satisfaction. As a result, 95% of our business today is based on recurring revenues. All of the largest law firms are our customers, be it the US-centric Am Law 100 or the Global 100 firms. In fact, we added the only missing Am Law 100 customer recently before I joined. We have an exceptional 95%-97% retention rate among our large law firm customers, which shows the depth of our relationship with these customers. At the other end of the spectrum, we have nearly 90,000 customers for our legal products and solutions, which also demonstrates the breadth of our industry presence. We truly believe that generative AI is a game-changer. It will expand our addressable market.
The legal industry is still a very, very labor-intensive industry and perhaps has not been as progressive as adopting technology as various other industries. With its strength in reading and writing, we believe that generative AI is especially tailored for the legal industry. You heard from David earlier about our vision for a ubiquitous AI legal assistant. And we are seeing unprecedented customer appetite for such tools. And we have products in the market already. So for us at Thomson Reuters, generative AI is both a vision and a reality. Now, with that background, I would like to discuss in more detail our existing business, the market trends that we see, and our differentiators and growth levers. So let's start with an overview of our current business. The Legal Professionals segment serves a diversified portfolio of customers and markets, with revenues fairly evenly distributed between these segments.
This includes the largest law firms in the world in our global large law segment, who have very complex and varied needs, and whose clients are some of the most important institutions and corporations in the world. We also serve the next tier of midsize law firms in the U.S. They are typically either focused on a geography or a practice area and tend to compete with the large firms in their particular area of specialization. Next, we see U.S. small law firms. These are firms ranging from one to 10 attorneys. They are more focused on consumer business and small business needs. Also, we have a large presence internationally. We have a particularly strong position in Canada and the U.K., where we are the market leader. We also have a broader presence globally beyond those markets, which my colleague, Matt Keen, will cover in a later presentation.
Finally, we have our government business that serves legal professionals that work in government institutions, including the judiciary, of course, but also the state and local institutions and federal agencies. We also serve risk professionals in the government space, including law enforcement agencies in particular. All our leading products, such as Westlaw, Practical Law, HighQ, and our AI solutions, have universal applicability across all these customer segments, which allows us to maximize the return from our product investments. We also have a few segment-specific solutions. This includes FindLaw, which is specifically focused on the small law firm market for business development, and our risk and fraud products, which are targeted at government agencies. Now, let's talk about some of the dynamics of the market. We have been benefiting from a set of trends that are driving increased technology adoption in the legal industry.
That is even before we start talking about GenAI. First, our customers operate in an increasingly complex legal and regulatory environment. Every year, we see thousands of pages added to the U.S. federal regulations. We see hundreds of thousands of new court cases and judgments in the U.S. alone. This is in addition to rapidly increasing corporate data. These increase our customers' need for technology solutions to help them clarify complexity and work more efficiently. At the same time, our customers face continuing and growing pressure from their customers, which are corporate general counsels, who demand that firms deliver work more efficiently and provide more value. Over the last decade, corporate legal departments have become more sophisticated buyers of legal services and are also more cost-conscious. They've introduced billing guidelines and shifted work away from more expensive firms.
Law firms need to earn the right and demonstrate to their clients that they're using the latest technology to be efficient in order to retain the work. Lastly, we're also seeing significant labor market shifts in the legal industry. That's driving greater tech adoption. About 19,000 new U.S. bar graduates enter law firms every year. These new associates are highly motivated to use technology to be effective and find work-life balance to be an important issue as well. They're really concerned about the technologies that their firms use. Simultaneously, we're also seeing a generational shift in the leadership levels as well of these law firms, with millennials now coming into these positions. Now, all of these trends are accelerating the use of technology in law firms.
These trends have been key to propelling our organic revenue growth from 3% in 2020 to 6%-7% in the second half of 2023. Now, let's talk about AI. With the advent of GenAI, we believe that the intensity and pace of technology adoption will only increase. If you look at the core work of lawyers today, it's mainly about complex language. It's about reading, writing, analyzing, synthesizing. This is why GenAI and language models hold a big promise for the legal industry. The pace of change is truly astonishing. You've probably heard that, unlike GPT-3.5, GPT-4 not only passed the bar exam but had a better score than 90% of those who took that. So it's not surprising to see the Goldman Sachs report identifying the legal industry as the number one industry in terms of potential for AI automation, with 44% of all work.
Our own very, very own research from the Thomson Reuters Institute has also found that 82% of customers truly believe in the potential to apply generative AI in legal. Our vision is that, in just a few years, we are heading towards a world where every lawyer will have a legal AI assistant, as you've seen before. A significant part of the work will be commenced by the AI assistant, for example, creating drafts of legal work but will be finished by the humans. In the medium to long term, as this technology gains prevalence, we expect to see a noticeable jump in efficiency and productivity of the legal industry to an extent never seen before. It's not just about productivity.
AI will also lead to higher-quality legal work, for example, the ability to review a more comprehensive set of documents and not just a sample at a given time, or also of precedents and recent regulations. We also believe AI will start to provide critical decision support, for example, on whether to litigate or not to litigate. Ultimately, we believe that it will free legal professionals from repetitive work and give them more time to focus on advising their clients and lead to more customer value and also higher job satisfaction. Finally, we see generative AI as a driving force for convergence between content and software. Today, law firms have siloed buying centers and organization structures for content and for software. Best-in-class usage of AI will require a more seamless technology architecture and buying patterns.
For example, personal productivity software, such as Word, for drafting contracts needs to be more tightly integrated with contractual clauses and Practical Law. Customer billing information in the future may need to reflect the work done by AI bots. Legal workflows and business development outreach may be triggered by new regulatory changes. In the midst of such changes and opportunity, we believe that TR is uniquely positioned to be a leader in the emerging AI-led transformation. Let me share with you the reasons why. The first reason TR is strong, as David described earlier, is that our authoritative legal content provides us with an incredible edge. Thomson Reuters has spent decades enhancing and structuring legal content. We also create proprietary content that can be found nowhere else, particularly in our market-leading Practical Law product.
These investments are especially important for retrieval-augmented generation, giving our AI-enabled products the highest level of accuracy and the highest level of relevance. Secondly, our domain expertise is unparalleled. We have over 1,400 highly skilled legal experts, in addition to 300 AI/ML practitioners, including talent who have joined us from the Casetext acquisition. The depth of the legal experts is not just important for creating the distinctive content we've already spoken about, but they also play an important role in training and fine-tuning our AI models to be even more effective and easy to use. The third reason why TR is strong is that we have an absolutely extensive customer footprint that we've spoken about, with the deepest relationships and the broadest presence across law firms and government organizations globally. This provides us a unique vantage point to effect change.
More importantly, we have a trust-based relationship that has been built with these customers over decades. Now, we all know that trust is a very important currency when it comes to AI. More tactically, we have an installed base of leading products, such as Westlaw and Practical Law, into which it's easier to introduce new AI features. And we have a large and experienced sales channel, which can easily be leveraged to cross-sell newer products, such as our legal AI assistant. Finally, I spoke about how AI is bringing together the convergence of content and software. And we believe that Thomson Reuters' content-enabled technology strategy, with the unique combination of authoritative legal content, workflow software, and AI capabilities, can be a winning play that will further differentiate us from competitors in this market.
So I hope that gives you a sense of why we are optimistic that TR can be a leader in this AI-based transformation of the legal industry. Now, a very important point that I want to emphasize is that AI is not only the future for TR, but it is here today. We have a head start. We have an early lead with three generative AI products already in the market. The momentum is building. Over 5,000 of our customers now have access to a TR GenAI-enabled product, which means one of Westlaw AI-Assisted Research, Ask Practical Law AI, or CoCounsel. Over 70 of the Am Law 200 firms have adopted one or more of our AI offerings already. We are expanding into other markets with at least one GenAI product now available in Australia and Canada, and as we heard from Jake in the UK starting today.
As you can see from the quotes here from our customers, we've received some great feedback on the power of our GenAI offerings they're already using. I'll give you a few seconds to digest them. My favorite is the one in the bottom right corner, where they feel that CoCounsel frees them to finally, finally do the work they went to law school for. And we are not stopping here. As you see in the discussion with David, we are launching a slew of additional AI products and skills later in the year, including our Microsoft Word-based drafting solution. And we are really, really excited about that. This is what I meant earlier by TR is already providing significant customer value with AI and is already monetizing it.
Now, let's spend a minute on how our addressable market is evolving as a result of these trends that we just discussed. The vendor market for legal technology solutions in areas where TR competes stands at $10 billion, growing by 6%-8% as current technologies get more widely adopted. We estimate a total addressable market of $23 billion, which includes $4 billion from GenAI. This is based on an early estimate for the addressable market for AI we see from the very first wave of productivity solutions that we've already launched. In other words, this is the potential that we see in the short to medium term. However, if we take a longer-term perspective, we may be erring on the conservative side. As AI capabilities continue to improve, we anticipate that customers will begin to drive more transformational changes, including in their operating model.
We could see a more pronounced shift from spend-on labor to spend-on technology. At a macroeconomic level, it's worth noting that U.S. law firms spend around $110 billion on labor today. Any material changes in this balance of spend between tech and labor will have a very big bearing on our addressable market in the next few years. So, with all of this background, let's review the key priorities that we see for the legal professional segments in Thomson Reuters. On the product side, we will maintain our leadership position in embedding AI into our existing products and launching new GenAI offerings across our product suite and across global markets. We will be laser-focused on the adoption and continued usage of our AI solutions so our customers receive significant and tangible value from AI.
We are also evolving our go-to-market by shifting from selling point solutions to selling complete productivity solutions and business transformation solutions. Lastly, we are developing new alliances and partnerships to expand our value proposition and influence in the legal industry. I believe if we execute and deliver successfully on these strategic priorities, we'll be in a strong position to capture the sizable addressable market we spoke about. So, let me close with a recap of our growth aspirations and our key drivers that will fuel our growth. We are targeting a continued improvement in our growth prospects for our legal professional segment, up from 3% a few years ago to 6%-7% currently, moving up to 7%-8% by 2026, as we play a leadership role in reshaping the legal industry. We believe this will be driven by three key factors.
First, growth in the addressable market with increasing adoption of current technologies and acceleration of new AI-based solutions. Second, TR capturing its fair share of the new addressable market emerging for AI-enabled productivity solutions with AI add-ons to existing TR products. And finally, TR playing in the newly converging area of content and software with a new legal AI assistant, with an increasing breadth of skills to create new revenue streams as well. So, finally, let me just say that I'm excited to join Thomson Reuters and lead the legal professionals team in this very pivotal time for the legal industry. Thank you very much. And I'd now like to introduce and welcome Elizabeth, the President of our Tax and Accounting Professionals Segment. Thank you.
Thanks, Raghu. Good morning, everyone. I'm Elizabeth Beastrom and President of our Tax and Accounting Professionals segment. I've been at Thomson Reuters almost 20 years, and specifically within tax for the last three. Tax, audit, and accounting, they're personal to me. I started my career at Deloitte as an audit consultant over 30 years ago. While the end result, whether it's an unqualified audit opinion or a completed tax return, has largely stayed the same, how you get there has evolved, enabled by increased efficiency utilizing technology. I love engaging with our customers every day. I am excited to share more about our business with you. We'll start by previewing four key messages I hope you take away from today. First key message: Tax and Accounting Professionals has a track record of delivering strong, consistent growth.
We have met or exceeded targets since 2021 while delivering approximately 10% organic revenue growth. We see a path to improve this trajectory. We operate in a large and growing segment with workflows ripe for automation. The industry is at the start of a meaningful transformation, heavily accelerated by technologies like cloud computing and artificial intelligence. I'll discuss more about the market opportunity in this presentation shortly. Third key message: The time is now. With an ongoing talent shortage, increasing regulatory complexity, and growing demand for services, tax and accounting firms are growing their technology investment to improve both efficiency and productivity. The fourth key message: Thomson Reuters is uniquely positioned to lead. We are investing heavily in our customers' future. Our leading products, embedded in our customers' workflow, combined with deep customer relationships, domain expertise, and proprietary content, provide a strong foundation for continued success.
We've made significant organic and inorganic investments, most recently acquiring SurePrep in 2023. We're investing even more over the next several years. The work our customers do really matters. They are trusted advisors to their clients and ensure tax and audit compliance for individuals and businesses of all sizes. We're proud to serve these firms with content-enabled software to empower them to efficiently provide the best services and outcomes for their clients. Our customers consistently tell us how much they rely on Thomson Reuters' products to do their work. This has enabled us to build a strong, stable, growing business with leading positions in the U.S. and Brazil, over 100,000 customers, including all of the top 100 U.S. CPA firms, as well as strong retention.
Facing growing demand and compliance-related complexity at the same time they navigate talent shortages, our customers need efficiency tools, which leaves their workflows ripe for automation and AI advancement. They're investing in technology, which fuels a $7 billion vendor market growing at 7%-9% annually. These conditions are ideal to power our formula for continued success in the tax professional segment: attractive end markets, leadership positions, a focus on leveraging our AI expertise to accelerate growth, and a deep understanding of our customers. Our tax professional segment is organized around our customers, with subsegments aligned by customer needs and centered around their workflow. Tax workflow is our largest subsegment and, as the name implies, is centered around tax services. Tax professionals want to drive efficiency, add value for clients, and provide a superior experience.
We help them with best-in-class compliance, AI-powered solutions, firm automation, and advisory services, with flagship products like Domínio, Checkpoint, UltraTax, SurePrep, and the full power of our CS Professional Suite. They buy from and stay with Thomson Reuters because we offer the most connected and reliable suite of products, supported by an open ecosystem that integrates with firm and third-party solutions. Our next subsegment is focused on audit workflow. Like tax professionals, auditors need efficiency and want to drive incremental value for their clients. Their workflow is increasingly digital, and harnessing the power of data and AI is unlocking new and exciting ways of working. We serve them with products like PPC, our industry-leading methodology, Cloud Audit Suite, and Confirmation. While tax by nature is highly localized, audit is a global workflow, enabling us to drive scale across geographies with these solutions.
Last, our strategic subsegment is comprised of the 30 largest firms, excluding the Global 7, which are served by our corporate segment. These firms have more complex organizations, broad service lines, enterprise tech stacks, a greater need for professional services, and different support models. We help them with connected software, content, and services across their firms to efficiently manage their business and deliver for clients across service lines. In addition to the solutions I mentioned earlier, we serve these firms with enterprise solutions like onesource, GoSystem Tax, and a host of workflow APIs. Across these subsegments, we have deep relationships enabling Thomson Reuters to be trusted partners in driving efficiency and enabling growth. We have a large recurring revenue base driven by the subscription nature of many of our products.
Our customers build their workflow around our solutions, and they tell me we are the backbone of their business, using our products to do their work, manage their firm, and communicate and work with their own clients. So we have high customer retention once implemented. Our customers must keep up with constant change. An ongoing talent shortage adds pressure, fueling a relentless drive for efficiency and automation. Firms are being squeezed on both sides. Experienced CPAs are exiting the profession in record numbers, while fewer new CPAs are entering, leading to a nearly 20% net decline since 2019. While this trend will eventually stabilize, it's unlikely to reverse. At the same time, firms are grappling with fewer CPAs, regulatory complexity, and demand for services continues to grow.
The pace of new tax laws, regulations, and audit standards isn't slowing down, and clients need help navigating issues like crypto and digital assets, R&D credits, and ESG, just to name a few. As a result, tax and accounting firms are focused on driving efficiency and automation. In fact, in our most recent Future Professionals survey, they told us now their number one priority followed closely by client service. They have accepted the reality of the talent shortage and want to augment their staff and help them to be more productive. Nearly 80% believe AI will help them achieve these goals. While talent challenges are arguably the most impactful trend in the industry, there are additional forces creating tailwinds for growth. The first is new technology, which is transforming the nature of the practice.
Many customers are still transitioning to the cloud, and when doing so, unlocking new efficiencies and improving how they collaborate and serve their clients. Paired with AI, longstanding first and last-mile challenges like document ingestion, data extraction, e-filing, and delivery can now be intelligently automated. And firms are hiring non-traditional talent from multiple disciplines to inject new expertise and skills to support new ways of working. Second, there's a continued shift to advisory services. Our customers want to be seen as trusted advisors, and their clients want ongoing guidance and advice from their accountants versus a once-a-year transactional relationship. Advisory is the fastest-growing service line in the industry and more profitable than traditional compliance work. However, firms are challenged with capacity and are investing in automating compliance workflows to free up time. Third, AI will be a major tool to drive this automation and supercharge tax, audit, and accounting professionals.
Our customers are overwhelmingly optimistic about AI's potential to boost productivity, with most planning to grow usage over the next three years. They also see adopting the latest tech like AI as a necessity to remain competitive. Finally, as use of cloud computing, AI, and other technologies grow, the need for connectivity grows. There's greater demand for products that work together seamlessly across workflows. As ecosystems emerge, APIs and data integration are critical to firm and client success. Our leading products embedded in our customers' workflow, combined with deep customer relationships and domain expertise, provide a strong foundation for continued success. We're going to accomplish this in several ways. First, driving increased efficiency by automating tasks across the workflow, dramatically reducing the time it takes for firms to deliver tax and audit work while improving accuracy and quality. Our customers tell us this is their greatest need.
For example, our 2023 SurePrep acquisition is achieving this with automations already in the market, like our industry-leading scan and populate capabilities, data classification, and real-time integration with GoSystem Tax. And we're continuing to invest in additional AI-powered automation that will roll out this year and beyond. We're investing to bring CoCounsel skills and new GenAI capabilities to our tax accounting and audit professionals' products to automate data ingestion, quickly compare and summarize data and documents, and streamline client communications. We're also bringing the power of GenAI to Checkpoint Edge to provide an improved and more personalized conversational research experience grounded by content our users know and trust. And these are just a small sample of the automations we're bringing to our customers. Next, we're enabling firms to deliver best client experiences. Through automation, we're freeing up time for staff to spend on higher-value client needs.
With GenAI that leverages our industry-leading content and guidance, we're empowering all levels of our clients' firms to deliver accurate, high-quality services. Lastly, we are well positioned to deliver for our clients. The TR AI platform enables us to rapidly deploy skills and functionality to automate, augment, and advise at scale. Our content and expertise enable us to quickly surface insights and answers in context, when, and where they're needed. An AIP-first approach ensures our solutions will work seamlessly together and with other firm solutions across the workflow. Finally, the deep and trusted relationships we have with our customers enable us to minimize disruption while implementing these new skills and capabilities, which is key to their success. This drive for efficiency, investment in automation, growing complexity, and demand for services is all fueling an attractive market opportunity.
Today, the segments we play in represent $7 billion of spend, with a total market opportunity of $16 billion. Transformational AI is increasing this market size, adding another $2 billion over the next five years, driving an $18 billion TAM. We expect our market to grow at 7%-9% annually over time. We're incredibly excited about the future for tax audit and accounting professionals and remain highly focused on executing against our top priorities: setting a new gold standard for modern, connected, intuitive solutions, delivering the most efficient AI-empowered workflows with end-to-end automation powered by data-driven insights, driving continued international growth, particularly from our strong Latin America franchise, and enabling an open, connected ecosystem integrating content and solutions that work seamlessly together.
We see a path to deliver continued double-digit revenue growth through 2026, with favorable tailwinds driven by increasing investment in technology enabling efficiency and automation, launching new AI-enabled products and features which command higher premiums and drive penetration, and expanding to new audit, advisory, and tax AI assistant product categories, adding new revenue streams across geographies. We target 2026 organic revenue growth of 10%-12% and have conviction in the opportunity and confidence in our ability to deliver. The tax audit and accounting space is going through a once-in-generation transformation, and with our longstanding industry positions, proprietary content, domain expertise, and AI know-how, Thomson Reuters is well positioned to continue to lead in the future. Thank you for your time, and I'd like to welcome Laura Clayton McDonnell, President of our corporate segment, to the stage.
All right, good morning. Thanks, Elizabeth. My name is Laura Clayton McDonnell, and I am the president of Corporates focused on supporting our enterprise customers ranging from the Fortune 100 to SMB. I have over 2five years of extensive global software sales and legal experience at leading companies in the high-tech field, including ServiceNow, Microsoft, IBM, and Apple. In my role, I am focused on customer and cultural transformation, driving positive results, growing market share, and developing high-performance teams. So let me start by quickly highlighting three key takeaways that I will expand upon throughout my presentation. First, we compete in a large and growing market. Corporates have the largest addressable market opportunity among TR's businesses, and it is also the least penetrated. Second, our corporate customers face several challenges that provide demand tailwinds for our business. Third, we are uniquely positioned to help our customers navigate these challenges.
We serve corporations across the globe in an attractive market segment, executing from a position of strength and aided by favorable market trends. We compete in large and growing markets that are driven by a number of favorable tailwinds, including rising regulatory complexity, new regulatory mandates, automation and digitization, and a focus on efficiency and productivity. These tailwinds play to our strengths, as our portfolio of trusted and authoritative content and compelling AI-powered software solutions are ideally positioned to help our customers navigate these trends. Our corporates business is focused on providing customers with products that serve tax, legal, and risk departments. We are a preferred provider of content and software to about 50,000 customers in all. Our business is primarily subscription-based, with high levels of recurring revenues and high retention rates.
Given the continued and increasing need by corporations for compliance and content solutions, our products have proven resilient through challenging macroeconomic periods. Our corporate segment is a $1.6 billion business, representing nearly 25% of Thomson Reuters' revenues. Tax departments in large corporations comprise 36% of our revenue, primarily driven by our direct tax and indirect tax solutions. Corporate legal departments make up 24% of our revenue, with risk solutions contributing 9%. 15% of our revenue is from global accounts, which encompasses both the Global 7 large accounting firms and another 33 multinational customers that we serve through a global account management model. International, including a contribution from the global accounts, is 18% of our revenue, and 85% of our revenue is recurring. Now, with the inclusion of Pagero, we expect our international revenue mix to rise into the low 20% range and our recurring revenue mix to improve slightly.
Next, I would like to elaborate further on what our customers are experiencing in today's market and how we help them. Now, our customers ranging from global multinationals to SMBs face a complex and challenging environment brought on by regulatory change and complexity, macroeconomic and geopolitical uncertainty, changing ways of work, and rapid technological advancement. Against this backdrop, they need trusted content and insights and powerful software solutions like those that we provide to navigate the many challenges they face. So I'll go into how our offerings help customers shortly, but let me first highlight three of these market dynamics that are providing a strong demand tailwind for our business. First, as Steve addressed earlier, our customers face a rising compliance burden driven by complex and ever-changing regulations that include new mandates, more onerous reporting requirements, and the potential for financial and/or reputational risk if not handled properly.
Examples of the increasing complexity include e-invoicing mandates, ESG reporting requirements, and global minimum tax regulations. Second, tax, legal, and compliance departments are all seen as cost centers and are under pressure to improve efficiency by doing more with less. For example, corporate legal departments are bringing routine work in-house, and corporate tax departments are leveraging automation to improve their efficiency and compliance. Third, corporations face rapid technology change, which, along with regulatory complexity, can create a complex web of data requirements, messy workflows, and inconsistent processes that result in the lack of transparency needed to support business decisions and operations across their enterprise. In addition, corporations are seeing increased pressure to understand and harness new technologies like generative AI. Well, these corporate challenges create significant opportunity for Thomson Reuters as we work to help our customers navigate the intersection of commerce and compliance.
In doing so, we have seen our customers generate new business, drive efficiencies, execute on their digital transformations, and improve customer loyalty using our technology. I find that the best way to illustrate how we help is through customer success stories. So I'm going to share two for you. First, a large technology company's tax team relied on us to digitally transform their indirect tax determination process. They were moving from spreadsheets with millions of data inputs to our ONESOURCE indirect tax determination offering, which streamlined their process and saved them more than $1 million in tax management costs. That was a huge savings for them. Other offerings like Practical Law provide similar efficiencies. Second, a manufacturing company with over 6,000 internationally sold products saw new revenue opportunities by automating their free trade agreement data management. Previously, manually tracking this data was inefficient and error-prone.
Implementing our ONESOURCE Free Trade Analyzer solution enabled them to streamline the process and identify new markets they were eligible to sell in. Well, these are just a handful of examples of how we enable our customers to succeed. We leverage the strength of our product for our customers' benefit every day, which I will showcase next. Our technology and insights deliver the visibility, the clarity, and the control that help our customers succeed. Our tax and trade portfolio reduces risk and regulatory complexity, allowing our customers' finance, trade, and tax teams to focus on driving growth. With ONESOURCE, we provide a one-stop solution for corporate tax and trade departments. Now, with the recent acquisition of Pagero, we are especially well positioned to be the end-to-end provider in the indirect tax workflow, spanning indirect tax determination, e-invoicing, and indirect tax compliance.
We are also excited about other applications of Pagero's Smart Business Network, and that is beyond e-invoicing. Our legal suite, it provides help to legal, compliance, and HR teams to work efficiently to protect their businesses, enabling them to provide better advice and insights and better manage their spend. Westlaw is the market standard for legal research, and Practical Law provides corporate legal teams with practical guidance, checklists, and market standard clauses. The launch of Westlaw AI and CoCounsel, an automated legal assistant, helps legal departments capitalize on the efficiencies of generative AI. Our Legal Tracker helps general counsel manage and track their spend on outside counsel and enforce spend discipline. Our risk suite allows customers to know with confidence if customers and partners are compliant and quickly investigate concerns. CLEAR is a premier solution that helps our customers conduct enhanced due diligence, investigate fraud, and comply with know-your-customer requirements.
Our competitive advantages are based on the expertise of our exceptional talent, our up-to-date content and powerful software, and the broad distribution footprint of TR. All of this is amplified by our extensive partnership ecosystem. Additionally, Thomson Reuters is investing more than $100 million annually on integrating AI into our portfolio, with an exciting roadmap of capabilities that David and his team reviewed earlier. All right, let me dig a little deeper into Pagero. As we discussed in January, we are extremely excited about the addition of this company and its wonderful team to Thomson Reuters. We see a significant growth opportunity in e-invoicing. We believe that Pagero has the market-leading solutions. We see a compelling strategic fit and admire several of its financial characteristics, including its highly recurring revenue and proven profitability in mature markets.
This strong fit of Pagero's e-invoicing capabilities with our ONESOURCE indirect tax offerings illustrates a number of the customer challenges that I discussed a few moments ago. First, Pagero solves a major customer regulatory compliance challenge through its e-invoicing offerings, which help companies comply with a wave of new digital tax regulations across more than 80 countries. Second, the workflow automation benefits of Pagero, especially when integrated with our ONESOURCE indirect tax offerings, bring both efficiency and accuracy improvement through digitization. In addition to e-invoicing, we are enthusiastic about several longer-term opportunities to leverage the Pagero Smart Business Network for additional automation and compliance use cases, including a combination with several other TR offerings like CLEAR and global trade management. We are excited to invest behind and drive deeper integration with Pagero as we execute the TR acquisition playbook to deliver long-term profitable growth.
The market dynamics that I have described bring significant opportunity for Thomson Reuters. We estimate that the addressable market is $42 billion, of which $9 billion is vended. We hold approximately a 50%-20% share of the vended market, which we believe is growing at 8%-11% annually. Relative to the other big three segments, corporates have the largest addressable market opportunity and is also the least vended. Now, this is in part because many corporations continue to utilize either inefficient manual processes, homegrown alternatives, or suboptimal solutions like spreadsheets. However, we see significant long-term growth potential for the vended market as the adoption of more automated and digital solutions, like many of those that we offer, continues to rise. Now, I would like to highlight our key strategic priorities.
We are well positioned to meet the needs of our corporate customers and are excited about our current and future opportunities. To take advantage of these opportunities, we will remain focused on four key priorities. First, we recognize the importance of creating an end-to-end customer experience across our go-to-market teams. Our value proposition, enterprise-level and account-based sales motion, and collaboration amongst our teams will enhance our customer experience. Second, in addition to our direct sales channel, we understand that our customers' ecosystem requires a close collaboration with a variety of partners. Our embedded models approach will ensure TR solutions continue to integrate with our customers' other preferred vendors or platforms. By doing this, we will be able to reach more customers. Third, with the regulatory environment likely to remain fluid, our customers will look to us to meet their needs.
We will continue to integrate our product suites to streamline their user experience. The recent acquisition of Pagero is a good example of this. Fourth, to make this all happen, we will work to make TR the preferred destination for world-class go-to-market product and operational talent, and we will continue to foster a culture of collaboration and growth mindset to address our customers' needs. I would like to close by highlighting four areas where our product and go-to-market investments are aligned with market needs. First, we will capitalize on the breadth of ONESOURCE to be the provider of choice as corporate tax and trade departments continue their automation journey. We will leverage the combined capabilities of ONESOURCE and Pagero's e-invoicing to provide a seamless experience to our customers. We will also drive additional use cases through Pagero's network to fully benefit from the growing network effect it provides.
Second, we recognize the preference of our customers to consume ONESOURCE capabilities pre-integrated into their preferred platforms like ERPs and e-commerce marketplaces. This helps solve multiple customer pain points and also allows us to tap into the large installed bases of third-party ecosystems. Third, our investments in introducing generative AI into Westlaw and Practical Law and the CoCounsel legal assistant, intelligent drafting, and legal spend management tools are perfectly aligned with the focus of legal departments to drive efficiencies.
Finally, with continued regulatory and customer interest in assessing counterparty risk, CLEAR and its associated capabilities, including the introduction of AI, will automate adverse media and sanctions screening, streamline workloads, and reduce false positives, in addition to delivering fraud prevention insights within the workflows. As we execute against our strategic priorities and deliver on these four growth opportunities, we are poised to compete and win, and we are targeting to deliver 8%-10% organic revenue growth by 2026. Thank you. And now I'd like to introduce Matt Keen, head of international, as our next speaker.
Thank you, Laura. Good morning, everyone. My name is Matthew Keen, and I'm the head of international at Thomson Reuters. I'm responsible for the performance and strategy for our big three segments in non-US markets, working closely with Raghu, Laura, and Elizabeth. I've been at Thomson Reuters for more than 20 years, holding a number of financial and operational leadership positions throughout the company. Today, I will discuss our international businesses and why we're excited about the opportunities our markets present. I'll start with four key messages I hope you take away from my presentation. First, our international businesses compete in large and growing markets. Second, international has been a key contributor to the big three revenue growth. And third, the theme of rising complexity that has been discussed throughout today's presentations is very much a global issue and provides demand tailwinds for our international businesses.
Fourth, we are focused on executing against three key growth drivers that I'll describe later. Before I take you through some numbers and the growth driver examples, I wanted to take a moment to tell you why international matters and what we are doing to deliver against the opportunity. So, what attracts us to our international markets? Our international market opportunity is significant at $15 billion, and the markets are growing at a healthy 8%-12% rate. Rising regulatory complexity and new regulatory mandates, like e-invoicing, which we just heard about, place a significant burden on our global clients, which is a key demand driver for many of our offerings. And we believe our international offerings are relatively under-penetrated versus the U.S., which provides significant long-term growth opportunity. So, what are we doing to make sure we take the opportunity in front of us?
We have a targeted international approach where we focus on several big bet markets in which we are investing both in distribution of our global products and also in select localizations and local market-focused products. These markets include the UK, Brazil, Canada, Australia, and Japan. Our broader international strategy is focused largely on our global offerings, which we sell both through our own teams and also through partnerships. As you heard from David, we are investing heavily in our product portfolio and bringing many of our innovations to key international markets. We see good opportunities to both add new customers and grow our existing relationships. Onto the numbers. Let's talk through our international footprint by segment, geography, and revenue type. First, it's important to note that the international revenue I'm discussing is the non-US revenue from our big three segments.
It does not include overseas revenue from Reuters or Global Print. From a financial reporting perspective, it rolls up into the big three segment results you've just seen. In 2023, international organic revenues grew 13%, reaching $1 billion, and represented 18% of big three total revenues. In line with the overall composition of our big three, legal professionals make up just under 50% of international revenues today. Corporates represent 32%, and tax and accounting professionals 20%. Corporates and tax and accounting professionals grew by double digits in 2023, while legal grew by high single digits. Our major products, such as Westlaw and Practical Law, make up a significant portion of our international revenues, with leading market positions in regions like Canada, the U.K., and Australia, serving both legal professionals and corporate customers.
ONESOURCE is the primary product within corporates, while Domínio, which is based in Brazil, is the largest driver in tax and accounting. I'll touch on Domínio in more detail momentarily. Moving to the right-hand side of the slide, from a geographical perspective, Europe is the largest region growing at high single digits. Drivers include the flagship products and legal and corporates that I mentioned earlier. LATAM is the second largest region in terms of revenue, which is mainly generated by local products in the tax sector. Domínio accounts for nearly half of the region's revenue. It has an established leadership position and additional opportunity for further penetration. AEM is our third biggest region and has been growing at double digits for the past few years.
Westlaw and Practical Law have been strong drivers of growth in AEM, along with corporate tax and trade and the recent acquisition of Westlaw Japan. Finally, Canada, which is 12% of international revenues and grew double digits in 2023, consists mainly of legal products like Westlaw and Practical Law, with small localized products and corporates and tax and accounting professionals also contributing. Lastly, looking at our revenue mix, it's attractive with 88% recurring revenues. As we look at the growth of our international businesses, we've experienced strong accelerating organic revenue growth in the last few years. This acceleration has outpaced the broader big three growth, resulting in international revenues increasing from 15% to 18% of big three revenues over the past five years.
This growth is a testament to the attractive nature of the international markets we serve and the targeted approach, along with our strong execution by our teams. Looking at the TAM, we estimate the target addressable market for our offerings to be $15 billion, with $7 billion of that currently vended. We also believe the market is growing between 8%-12%. By geography, the big bet markets I mentioned earlier represent the largest opportunity. As Laura just mentioned, the corporate segment addressable market is also the largest opportunity for international, with sustained double-digit growth due to rising regulatory complexity and the globally scalable nature of our offerings, for example, in e-invoicing, indirect tax, and global trade management. Now, to the first of our three growth drivers. Throughout the day, you've heard and seen some of the great things we are doing with generative AI across our product portfolio.
What I'm showing here is how those investments are being taken to market internationally for our legal customers, across legal professionals and corporates, building on top of our successful product releases here in the U.S. As Jake mentioned, we've already launched CoCounsel Core in Canada and Australia, and today we launched it in the U.K., with further releases in other markets due to follow rapidly on the back of that throughout this year. In addition to CoCounsel, I'm also really excited to see Westlaw AI-Assisted Research and Ask Practical Law AI and our intelligent drafting solutions coming soon to the U.K., Canada, and Australia. The second key driver of growth for international is Domínio, a leading provider of accounting software in Brazil.
With a strong focus on compliance, Domínio has positioned itself as a trusted solution for businesses of all sizes, including small firms, which represent 75% of the Brazilian market. Domínio's success is driven by its ability to deliver a comprehensive and constantly updated solution that ensures compliance with Brazil's highly complex tax and payroll regulations. Domínio supports over 3 million small and medium businesses through its 40,000 customers, boasting a 43% market share, and it continues to grow quickly. It is highly embedded into tax firms' workflows and offers exceptional customer support, evidenced by a 92% retention rate and net promoter score of 71. Since its acquisition in 2014, Domínio has grown at a compound annual growth rate in excess of 20% and has been a key growth driver for international and tax and accounting professionals.
One last note on Domínio: in Kirsty's discussion of internal GenAI opportunities earlier, there was a mention around use in customer success teams. We've applied that with our teams in Brazil and already seen improved customer experience and faster response times. Back in January, you heard from Steve, Mike, and Gary talk about our acquisition of Pagero. You've just heard Laura talking about it now. I won't talk about everything that Pagero offers our customers through its smart business network. I do want to use Pagero to illustrate how regulation and mandates are changing in a way that we can benefit from. If I'd shown you this picture +five years ago, there would have been a very small number of countries shaded green. Fast forward to now, and more than 80 have announced or implemented e-invoicing and digital tax reporting mandates and more are likely.
This rapid growth has contributed to Pagero's 25% annual growth rate in recent years. I'd also like to note that more than 90% of Pagero's revenue comes from outside North America. So, we're really excited to bring this solution to our global customer base to help them navigate the increasingly complex web of digital tax regulations. Before I hand you over to Mike Eastwood, our CFO, let's quickly recap. International markets are attractive and growing quickly.
We have a track record of performing well with consistent double-digit growth over the last three years and a favorable tailwind going forward. We talk through three key growth drivers: taking our GenAI products to international markets; our strong Domínio business in Brazil; and Pagero in the indirect tax e-invoicing opportunity. I hope these three growth driver examples give you a flavor of why we are confident in our outlook for the big three international organic revenue, which we target growing at a rate of 13%-15%. Thank you. Now, let me turn it over to our Chief Financial Officer, Mike Eastwood.
Right. Thank you, Matt. And thank you to those in person and listening through the webcast today. I hope you have found today helpful. As you can tell, we are excited about how our company is positioned and the many opportunities in front of us. I hope you also appreciate our talented team who have presented today. I will echo Steve's earlier comments around the strength of our team, which we see as a real competitive advantage for our company. With all of the exciting stuff out of the way, let me bring the day home with our financial review. I will start by highlighting four key messages I hope you take away from my presentation. First, we have delivered against our 2021 targets while becoming a stronger company. Second, we see a number of attractive organic and inorganic investment opportunities, and we are investing heavily in 2024.
Third, we are confident these investments will pay off through faster revenue growth and a return to year-over-year organic margin expansion in 2025 and 2026. Fourth, we have significant capital capacity, and we will remain disciplined in how we deploy it to create long-term shareholder value. Let me start with a recap of our performance from 2021 to 2023. Steve and Kirsty both discussed the success of the Change Program. David credited our technology infrastructure improvements as a key factor in the rapid development and launch of generative AI capabilities within Westlaw Precision in 2023. Let me provide a bit more color on our financial performance. In February of 2021, we laid out three years of detailed financial targets that called for accelerating revenue growth, rising margins, and significant free cash flow growth.
We also stated our intention to begin monetizing our stake in the London Stock Exchange Group and discussed our approach to putting our capital capacity to work. I am happy to report we have delivered on all fronts. I will expand upon these achievements over the next few slides. Our organic revenue growth has tripled in recent years to 6%, from 2% in the decade through 2020. Our big three segments have also strengthened, growing 7% in 2023, up from 4%-5% in 2019 to 2020. There are several factors that have contributed to this improvement. I will briefly mention two in the next few slides. First, in 2021, we highlighted the seven key growth initiatives shown here and stated we would increasingly prioritize investment of both management time and financial resources behind what we believed were among our strongest franchises and best growth opportunities.
Those efforts have paid off with strong revenue acceleration from these key offerings. We feel very good about the outlook for these franchises, which we expect to continue contributing to revenue growth acceleration over the next few years. Second, our portfolio optimization work has also contributed to our improved revenue growth and outlook. We continuously assess our portfolio for strategic and financial fit and to consider if we are the best owner for an asset. This work resulted in several small non-core divestitures in 2022 and the more meaningful divestiture of a majority stake in Elite in 2023. In aggregate, we have divested approximately $350 million of revenue that was not growing. At the same time, we have completed nine acquisitions since 2021 for $2.2 billion, which have brought approximately $200 million of revenue growing in excess of 20%.
In total, this portfolio optimization work leaves us with a more strategically aligned portfolio that has stronger growth prospects. I will now shift to profitability, where our Change Program efforts, along with improved revenue growth, have driven a significant improvement in margins and free cash flow. Reducing complexity and the transition to an operating company have been meaningful contributors to the 600 basis points increase in our margins since 2020. For example, we have streamlined our product portfolio, halved our office count, and meaningfully increased the mix of our workforce and our global centers. These factors have also contributed to the sharp increase in our free cash flow, which has occurred despite heavier than initially expected capital spending. We have also made strong progress at monetizing our LSEG stake, with nearly $5.5 billion sold in 2023 and another $1.1 billion last week.
We currently own 5.9 million shares worth approximately $700 million. In total, we have monetized $7.6 billion, or 93%, of our initial stake. The return to Thomson Reuters and our shareholders has been significant. When we sold a majority stake of our financial and risk business to Blackstone in 2018, our minority equity stake was valued at $3 billion. The stake grew to $6.7 billion in value when Blackstone and Thomson Reuters agreed to sell Refinitiv to the London Stock Exchange Group in 2019. Today, between our LSEG monetization to date and the value of our remaining stake, this position is worth $8.3 billion, or nearly triple the initial value in just over five years. I will wrap up my 2021 to 2023 lookback by repeating a slide that Steve showed earlier.
Over the last three years, we have consistently delivered to the targets we provided in February of 2021, achieving or exceeding our outlooks for revenue growth, profitability, and free cash flow throughout this period. For capital intensity, we made a deliberate decision to invest more in response to the many opportunities we see across our business. This higher investment is largely targeted at generative AI and our recent acquisitions, which are expected to be key drivers of the revenue acceleration we forecast in 2025 and 2026. I will now discuss our 2024 to 2026 outlook. On our fourth quarter earnings call in February, we provided the 2024 guidance shown here. We forecast organic revenue growth of approximately 6%. We expect the big three segments to grow approximately 7.5%, continuing the strong trend of modest acceleration we have seen in recent years.
We are forecasting a 2024 adjusted EBITDA margin of approximately 38%, down from 39% in 2023, with the modest margin decline driven by our recent M&A and higher organic investments. As we had previewed earlier in 2023, we are choosing to reinvest our underlying operating leverage in 2024 into accelerated organic investments, particularly in the generative AI area. We do not take this decision lightly, but we see significant opportunity through these investments to expand our medium- to long-term growth profile. I will provide a bit of incremental color on both our inorganic and organic investments. With our Q4 earnings, we stated M&A activity since mid-2023 is expected to be roughly 120 basis points dilutive to 2024 margins. Including 35 basis points of integration expenses, we expect to fall off within 24 months.
We are bullish on these businesses and believe the initial margin pressure is a sound investment that will pay off handsomely over time. We expect approximately $240 million of revenue from these acquisitions this year and believe it will grow to $650 million by 2028, a nearly 30% compounded growth rate over the next few years. This outlook is based on our expectations for underlying growth driving cross-sell through TR's distribution and new product innovation through product integration and continued investments in the acquired teams. We also believe this portfolio of businesses we see their profitability scale strongly with revenue, approaching TR-level margins over the long term. In addition to our inorganic investments, we are raising our organic investment levels in 2024 to position the company for improving revenue growth.
Our investments are focused on several areas highlighted on this slide. First, we are investing heavily in product innovation, as David discussed earlier. These investments are heavily focused on generative AI but also include a number of other product initiatives. Second, as Kirsty discussed, we are investing in infrastructure, including our Skills Factory platform that will help us scale and accelerate our GenAI innovation in the future. Third, we continue to invest in our go-to-market and customer experience initiatives, including partnerships which Laura discussed. We are confident these investments, along with our inorganic investments in M&A, will contribute to organic revenue growth improvement going forward. Looking beyond 2024, we provided the 2025 to 2026 financial framework shown here on our fourth quarter earnings call.
In this period, we target an organic revenue growth range of 6.5%-8%, driven by 8%-9% for the big three segments. We will work to deliver acceleration within this range over the next few years as we expect our GenAI investments to pay off and recent M&A to scale. For margins, we anticipate delivering 75 basis points of expansion in 2025, followed by at least 50 basis points annually thereafter. I would note this is an organic outlook and could be impacted by future M&A. We expect our capital intensity to remain at approximately 8%, or relatively stable with the recent trend, after some of our acquisition integration spending moderates. Lastly, we expect our free cash flow to remain robust over the next several years, growing to a range of $2 billion-$2.1 billion in 2026.
This assumes some further increase in our effective and cash tax rates beyond 2024, stable capital spending, and rising margins. Let me provide some additional color on our revenue and margin drivers. We see several key drivers of the targeted 2025 to 2026 acceleration. First, we expect our recent M&A to be accretive to our organic revenue growth. For 2024, we anticipate roughly 50 basis points of benefit, with that benefit growing in 2025 and beyond as the acquired businesses scale. As I stated earlier, we expect $240 million of revenue this year from our 2023 and 2024 acquisitions and see this base growing strongly into the future. Second, we are bullish on the momentum in our base business and the strong customer interest in our robust product pipeline. As David discussed earlier, the pace of our innovation has accelerated meaningfully in recent years.
As Steve and Raghu discussed, the momentum and Westlaw, our largest franchise, has picked up meaningfully following the introduction of Westlaw Precision and the more recent addition of the AI-assisted research capability. We see 50-100 basis points of acceleration from each of our recent M&A and our product pipeline, including generative AI offerings. Tempering these benefits slightly, the sharp devaluation of the Argentine peso will have a modest negative impact on our organic growth, as I discussed during our fourth quarter call. Our confidence in delivering the 6.5%-8% range is high based on both our M&A and organic product momentum, along with potential for revenue retention and pricing improvement. I discussed earlier the benefits from our portfolio optimization since 2021, and Steve discussed our current revenue momentum and the base of products growing at double-digit rates.
This slide combines these concepts, illustrating the improvement in our revenue mix since 2019, with the base of double-digit growers nearly doubling while our Global Print mix has declined. Including the annualized benefit from our recent M&A, particularly Pagero, we expect the mix of double-digit growth offerings to increase to the low 20% level. As these products continue to scale and our mix of print declines, our improving revenue mix should contribute to growth acceleration going forward. In addition to revenue acceleration, we are confident in our ability to return to year-over-year margin expansion in 2025 and 2026, driven by several factors. Our highly recurring revenue and 60%-65% fixed cost base drives healthy operating leverage over time. We have a continuous improvement philosophy and are always focused on ways to drive incremental efficiency.
Continued migration of our team to our global centers and leveraging internal GenAI opportunities are two potential levers, as Kirsty discussed earlier. As our efforts to improve the customer experience continue, we see room to increase our revenue retention above 91% over time. We expect the profitability of our recent M&A to improve as we scale revenue in the next few years, especially in the third year when our integration expenses will have largely fallen off. Against these positives, it is worth noting incremental M&A tends to bring lower initial margins. We understand we are investing significantly in 2024, and we feel a responsibility to deliver a return on that investment to our shareholders in the form of both faster revenue and a return to margin expansion. I will finish with a few thoughts on our capital allocation approach and history.
As you have heard me explain in the past, we follow a balanced approach to capital allocation. We prioritize organic investment and dividend growth, followed by strategic M&A and capital returns. Since 2021, we have allocated capital across these priorities, including the following. We have grown our dividend 10% annually in each of the last three years and paid out common share dividends of $2.5 billion since 2021. We have increased our dividend for 31 consecutive years. We have invested $2.2 billion into strategic acquisitions since 2021, bolstering each of our big three segments along with Reuters. We have returned $5.8 billion to shareholders through share repurchases and the 2023 return of capital transaction, retiring nearly 11% of our shares outstanding since 2020. Looking forward, we remain extremely well-capitalized with a net leverage ratio of approximately 1x and strong free cash flow.
As Steve mentioned earlier, we estimate capital capacity of approximately $8 billion through 2026, which we remain focused on putting to work in a way that maximizes long-term shareholder value creation. M&A remains a priority, and we have had good success, closing 6 strategic acquisitions since the beginning of 2023. We remain squarely focused on executing what we call the Thomson Reuters Acquisition Playbook. This entails acquiring high-quality businesses within our areas of expertise, integrating them into TR's product suite, investing behind their growth, and leveraging our extensive distribution and customer relationships to drive profitable long-term growth. Our track record in this regard is strong, and we believe SurePrep, Casetext, and Pagero will benefit from this approach. Let me briefly comment on our approach to M&A.
We follow a disciplined process with potential acquisition targets needing to bring a compelling strategic fit with our portfolio, meet our financial hurdles, have strong cultural fit, and bring modern technology. Financially, our analysis is grounded by a 10-year IRR NPV framework, which we use as the basis for all M&A transactions. We target an IRR of at least 2x our weighted average cost of capital and consider a number of additional metrics, including payback period, integration complexity, return on invested capital, organic growth impact, and accretion-dilution to free cash flow and margins. We also risk-adjust this analysis based on the characteristics of the particular transaction being considered. It is worth noting we use this framework not only for M&A but also to analyze the attractiveness of share repurchases and large internal capital projects.
This financial rigor helps us remain focused on shareholder value creation and finding the best uses of our capital capacity. To this point, we acknowledge share repurchases are less attractive today, with interest rates around 5% as compared to a year or two ago. As a result, we see strategic M&A that meets our disciplined approach as more attractive than incremental buybacks in the short term. Our balance sheet remains a key strength with a year-end net leverage ratio of 0.8 times. Pro forma for the Pagero acquisition and the recent LSEG transaction, our leverage ratio remains approximately 1 times. We have significant borrowing capacity through our cash on hand, undrawn $2 billion revolving credit facility, and approximately $1 billion of availability on our $2 billion commercial program, in addition to our healthy free cash flow. On our fourth quarter call, we introduced the following capital strategy targets.
In addition to maintaining the prior 2.5 times leverage target and 50%-60% dividend payout ratio, we have added two new targets. We target returning 75% of our annual free cash flow through dividends and share repurchases, and we target a return on invested capital that is at least twice our weighted average cost of capital. I will close my prepared remarks by sharing our value creation model, which has been tweaked slightly to incorporate our new 2025 to 2026 organic revenue framework and our 75% capital return target.
This model continues to guide our long-term investment approach. As we execute to its principles, we believe Thomson Reuters is positioned to consistently and sustainably drive strong operating and financial performance that builds value for our shareholders over the long term. Let me now welcome Steve and our segment leaders back to the stage and hand it to Gary to moderate a final Q&A session.
How about we start with Vince back here?
Thanks, Gary. Vince Valentini from TD Cowen. I'm wondering, just on that last slide and all the capital you have to deploy, can you just level set us on I mean, it's a fantastic story and gives you a runway of long-term growth, but can you level set us on what that means for margins and free cash flow? If you spend a lot of that money on acquisitions, you kind of admit that'll hold back the margins temporarily. If you shift more to building more of the capabilities in-house, that probably drives your CapEx or keeps it higher, which also impacts free cash flow. So is it just a necessary evil that we should probably see a little bit less free cash flow conversion as you see this acceleration towards maybe even double-digit revenue growth beyond 2026?
Yeah. I'll start, Vince. First and foremost, we are committed to delivering our 2024-2026 targets. What I was trying to indicate today is optionality could influence the sequencing and timing thereof, but as we sit today, we're fully confident in delivering on 2025 and 2026. Certainly, to your point, and as I mentioned, M&A has a tendency to dilute margins in the near term as we accelerate over the long term. Your point in regards to organic investments, certainly that could impact capital intensity, but we go through a very rigorous process to prioritize those. Key message today, based on what we know, we're committed to delivering that free cash flow and margin expansion in 2025 and 2026.
Just to add to that, Vince, one of the things that Kirsty and her team are laser-focused on is constantly grinding down our spend on KTLO, right, on sort of addressing legacy tech debt and rolling through service health remediations and product fixes. That's gruesomely hard work, for sure, but we're making progress. To your second point, where we're making progress is to take, as a proportion of our CapEx, is to take that down each year, year-over-year, and be able to spend that increment, or at least a large part of that increment, on more R&D.
And the journey we're on is, particularly for David Wong and Shawn Malhotra and their teams, our Head of Engineering, David, obviously our Chief Product Officer, is to really sort of prove our mettle in terms of R&D and new product innovation, and at the same time, have Kirsty reduce that KTLO and be able to spend more and more. So I think the second part of your question, if we're successful, we will not be sort of ratcheting that up to drive greater organic innovation. And on the first part of your question to M&A, as we see M&A, obviously, we're proud of the track record we've had.
We're thrilled with the acquisitions we've made, particularly over the last sort of 15-18 months. It's very hard to predict what we'll be able to get done, but I will promise you one thing, and that is we will remain incredibly rigorous about this. We will not get deal fever just because we have this $8 billion sitting there. If we can find the right deals, we'll do them. If we can't find the right deals, we'll keep moving. And so we're not going to sort of move away from that framework, which we think has got us in good stead so far.
How about right next to Vince? Bobby?
Hi. Bobby Reynolds with Fidelity Canada. Could you talk about what could drive upside to the 6.5%-8% range, either within the three years or beyond that? And I'm curious, the discussion about legal assistance potentially being on every lawyer's desk within the next three to five years, is that incorporated in that range, or would that be sort of upside to think about beyond the three years? Thanks.
Yeah. Happy to start, Bobby. I'll mention three or four items. I think the pace acceleration sequencing of the GenAI adoption is a key factor for us. I think we've consistently stated that we expect the contributions from GenAI to pick up in the second semester of 2024 and into 2025. So that's the first item that could provide some additional opportunity. Second item is retention. Our weighted average retention based on revenue is 91%. We've had small incremental improvements, call it 20-ish type basis points over the last few years of annual improvement. But if you compare that, and I think we've talked in the past, Bobby, to some of our business information services, peer group, best in class, maybe in that 94-ish% range, can we get to 94% during that time horizon? That would be a tall task.
But based on what we've discussed today, we have a lot of optimism in regards to the ability to increase our retention, especially as we continue to improve our net promoter score. Net promoter score, as Steve and Kirsty mentioned, not where we want it. I think there's a direct correlation in the improvement of NPS and the improvement in retention. Third item I would mention, Bobby, is the recent M&A. We've factored in the M&A. It's early days yet. Might there be some opportunity over the time horizon with those? Possibly. That's one that we just need more time. And then the fourth item I think I missed in the prepared remarks was on pricing. That's something that has creeped up just incrementally in recent years. Those would be the four items that I would say on the radar. Steve, any items to supplement?
No, well said, Mike.
You've got Tony and also Drew.
Drew. Yeah.
Manav, yeah.
You can go. Yeah, you can go here.
Yeah. Thanks very much. Drew McReynolds from RBC. Two for me. First, David talked about the pace of innovation accelerating. I think for analysts that have covered you for a long time, we actually see that, and you did that in 2023. Just wondering how you're feeling relative to the competitors out there in each of the different segments. And then secondly, a question I wanted to ask this morning, so I apologize if it's, again, a David question, but just wondering
On the reinforcement learning of the AI assistants, it feels to me that once it gets going, creates a pretty interesting moat in terms of why would you switch off an assistant that is essentially kind of personalized or company-specific? It just seems that that's a pretty good kind of new moat that forms in this whole kind of ecosystem. So I'm wondering if that assumption is correct. Are you able to do reinforcement learning today by individual, or is it kind of the domain at the company level?
Yeah. So let me comment in reverse order. So yes, I think, and Jake, you should feel free to disagree with me here because you've been living this for a while, but I think once a professional becomes accustomed to using an assistant, and that assistant is increasingly tuned to that professional's usage patterns, the switching costs, for sure, Drew, go up. And so that's why we're very focused on driving adoption and the sort of after-sales support as we go through. And so I agree with that premise. I mean, it's early days, but I think that's certainly something that we're focused on. In terms of the competition, and I'll ask the presidents if they care to add to this, it's a delicate balance. I mean, we are very respectful of our competitors. We have some terrific we compete with some terrific companies and great folks therein.
However, we do think that if we just continue on this path of getting better and better at understanding our customers' needs and being able to translate those back into our products and deliver on that innovation roadmap, that we ought to be able to further the distance between us and the competitors, particularly the folks who we've been competing with for a number of decades. We're always asked the question, "Well, what about the sort of two folks in a garage and the startup who've got some VC money?" And there's obviously been a lot of VC money coming behind generative AI. I think what we're learning is that the existing customer relationships we have are incredibly important. I mean, Mike Dahn's product is in the hands of about, I think, Mike, 5,000 practicing attorneys now a couple of months after its launch.
And so being able to sort of use that distribution mechanism, those relationships, that intel is important. Secondly, the content is king in this sort of first, and we think, first couple of chapters of generative AI. We have it, and most others don't. And those that do have it don't have it in the same quality that we have. And we're going to keep that investment going. So we're not going to blink when it comes to continuing to invest behind our content and editorial teams. But I'd ask the presidents to add anything on that on the competitive front. Laura, do you want to go first?
Oh, sure. I'll go first. I mean, just to supplement what Steve just sort of disclosed there, I mean, I think that the core of it for us is when you have TR as that preferred partner, it's because of the trusted content. It's authoritative. It's there and has been in use with those customers for quite a long time. And so they have developed a reliance on Thomson Reuters to be able to provide there, to be there not only today but into the future, to continue to inform the way that they do business. So I think that separates us a little bit. And so we try very hard to hold that dear and keep that at the center of everything that we do.
Yeah. And just to add on to that, I mean, we have deep customer relationships. And with that is understanding what our customers need. And what our customers need is automation, efficiency, and they need to solve for the talent shortage and do a shift to advisory. And we're well-positioned from our current product roadmap as well as future product roadmap and innovation as well as inorganic options to meet that need, so.
Yeah. With regards to the legal professional segment, I'm new in this role, so I don't feel yet proficient to comment about competitors and rather focus on TR. What I see is it's a brand new market. A lot of things are transforming. We had the heritage. We had the leadership. I see this as a market to approach with a challenger mindset. It's a market share game. It's an early day. We have, of course, strengths and content which play to us, but it's an opportunity. It's almost a new opportunity for us to win.
Just think internationally. If you think about the legal product suite, you look at what's coming and what's coming from the product teams you've seen earlier. I think that just makes my team really excited about what they can do in the market. And I think if we talk about corporates, Laura touched on, sometimes our biggest competitor is the guy doing it himself internally and explaining and helping them on that automation journey and adopting our tools. We can really bring value to them. And then once we've shown that value to them, we can go further in their organization. So I think I'm just very excited about what we can do with the suite that we have available to us in all the markets that we serve, particularly for global customers.
Maybe just to jump in, one last thought on the reinforcement learning question. One, it's still very early. And so the labs' teams, Kirsty's teams, Jake's teams are all doing experiments across different techniques, as I mentioned, around either training new models, refining existing models, or applying off-the-shelf models with the content. We hypothesize that our expertise and the experience that our practitioners have will help with that reinforcement learning. Some examples we've already seen are actually somewhat counterintuitive, like teaching the system to say, "I don't know." That's actually super powerful, where fine-tuning the system so that it says, "I will only respond based off of what I see in my database," is a way of engendering trust with our customers.
And that's an example of where we've seen our researchers being able to help to create a system which responds more reliably because you don't want it to hallucinate. You actually want it to tell you where, "I don't have the informat on." And that's something with ChatGPT, as an example, doesn't do very well because it just wants to answer your question. On competitors, one last thing I'd share is we are very respectful of our competitors. And we recognize that the pace is going to go up in the marketplace.
And so we see competitive moves as a way to learn what's happening in the marketplace. And things that we're doing, such as staying closer to developments in the startup ecosystems with TR Ventures, is an opportunity to learn and to stay close to the pulse of what's happening. All that being said, one general trend we see with our customers is that they are a little risk-averse. And so they are looking for the right solution to solve the problem for them. They're willing to experiment, but they'll put their money behind the solution that they trust.
Should we come maybe up front to Tony?
Thanks. Tony Kaplan from Morgan Stanley. Steve, you mentioned earlier in the presentation, and forgive me if I mischaracterize, but creating a GenAI assistant for all professionals outside of legal and I understand how you're doing it within legal and tax given the really vast content sets that you have. And I know you mentioned maybe bringing in other content as well. I guess, how do you think about starting that? It sounds ambitious unless you're targeted. And so how does that sort of play out? How much investment? How meaningful can it be? Anything around sort of that opportunity because it sounds.
Yeah. I'm going to start. Then I'm going to ask Jake to supplement. In the first meeting that Jake and I had at a restaurant in Silicon Valley, I asked him a question, which is, "Jake, if we could give you 170 years or let's shorten it up, 50 years of the Reuters news file, what would you do with that? And how would CoCounsel interact with that?" So I certainly have had in my mind that what Jake has built, what he and his team have built, is first and foremost relevant to lawyers because that's who they principally built it for. But I think it's very powerful for tax and accounting professionals. I think ultimately, it'll be powerful quite soon for news organizations.
I also think it'd be powerful for risk, fraud, and compliance and ESG, and those areas that we sort of play in to varying degrees. Beyond that, would we ever sort of be serving medical professionals? I don't think we need to, candidly. But certainly, my view is our acquisition of Casetext was, yes, about legal, but it was about much more, Tony. But this may be one of those examples where the CEO is in one place and those who actually know what's going on are in a different place. So I'll defer to Jake.
That's right on. The only thing I'd add, because I agree with everything Steve said. The only thing I'd add is that we call it generative AI because a lot of people are focusing on the fact that they can generate stuff. But to me, that's actually the least impressive piece of it. The most impressive piece of it is these modern models think, understand, read, and to some degree, logic at the level of a, say, postgraduate. We were the folks who did that original study when we got early access to GPT-4 that showed that when it took the bar, it didn't just pass the bar. It did better than nine out of 10 bar test takers, people taking the bar. And that's just where we are with version four, what happens at 4.5 and 5, etc., and other companies' models.
And so when you have a technology that for the first time can operate that kind of level of intelligence, the applications to assist professionals, broadly speaking, and especially, of course, the professionals that we serve at Thomson Reuters, explodes exponentially. The amount of things that we were able to do for our clients, even as a small company before being integrated into TR, was already pretty large. And now we have 10x that opportunity. So that's something we're really excited about.
Manav?
Manav and then Andrew?
Yeah.
Thank you. Two questions. The first one, for your target revenue growth rates in legal and corporate, you have them growing kind of in line with the market. I'm not complaining about the growth rates generally, but why not above the market with all the excitement you've talked about compared to the other divisions? And then the second part, just on international, I think the story makes sense. But what is the limiting factor? Do you need the content internationally to make it a bigger piece? Do you need to acquire more companies like Domínio, or why isn't that a bigger mix already?
Manav, I'll start on the first question in regards to legal. I think that was one of the points. I think corporates Raghu talked today about 7%-8% organic growth by 2026, three items that we have a close eye on that we're monitoring. One is our FindLaw business. I've mentioned it during our last two earnings calls. It's about $300 million of annual revenue in the legal business that has been flattish over the last few quarters in full year 2023.
Second item is our government business, which is roughly 6-6.5%. We're working on opportunities to scale it. We've talked about it during several earnings calls during 2023 with it. Then the third item, which goes back to Bobby's question, what could create potential upside, is the pace, acceleration, sequencing of the GenAI adoption. We just don't know enough yet to fine-tune that. We'll go to corporate in a second, but maybe, Raghu, your thoughts.
Yeah. Just to add to that, I think it's about expectations on the adoption of the legal AI assistant and the pace at which it'll go. I think I told you that we have 5,000 customers of our 90,000 using some kind of AI products. As you would expect, these are early adopters, right? These are the first ones, easier to get. The question is when the mainstream will come online. And I guess we are being prudent about it to witness it before we can actually capitalize on that.
Maybe the second part of the question, I'll tee it up and pass it to Laura. Corporate, 7% organic growth in calendar year 2023. Laura mentioned 8%-10% by 2026. Several items that are going to impact the flow there is the Pagero acquisition, which we talked about both in Laura's presentation, I think, and Matt's. A second key factor for us in corporates, I think it was 90% retention rate. We just had incremental improvements there. The pace of acceleration in net promoter score and retention will be really important for us in corporates. And then the third item I would mention is cross-sell. I mentioned cross-sell during my prepared remarks. But those are three items, Laura, that I would mention that we're watching really closely that could influence the pace and acceleration of sustained organic growth. But your thoughts, Laura?
That's very good, Mike. I think I would add to it is that the other part of the equation is customers making a decision to automate. And I think there's a little bit of delay, right, because there's uncertainty with the macroeconomic environment. There's complexity out there. They're trying to figure out, "Well, is what I'm doing today, is that just good enough?" And I think ultimately, corporations and customers overall are going to come to a point where they're going to have to make a decision because there's a scarcity of skilled workforce, right? And so when you have that, there's almost going to be this choice ultimately that a corporation's going to have to make. And that is to automate and to innovate.
When you think about that, I think one of the decisions that they're going to make is, "Well, who are the providers out there to make that happen?" Thomson Reuters is one of them. So as Mike said, Pagero helps us get within the range because it's a great opportunity. But as tax and trade departments think about how do they automate their journey? How do they provide an end-to-end experience? They're going to look to our ONESOURCE product portfolio. Customers are also looking at their ERP systems and e-commerce systems. And ONESOURCE is a way to sort of embed it there. Then when you think about adding e-invoicing and compliance, now you've got this end-to-end compliance solution. Of course, we've talked about the work that we're doing and investing in generative AI with Westlaw and Practical Law.
I'll just round this up by CLEAR. When you think about how we're moving across the whole of our portfolio and integrating generative AI, CLEAR is a prime opportunity to do that too. So for corporates, it's a little more complex. But in some, I think there's a little bit of delay. They're thinking about what's happening in the economy. They're trying to make the right decision for their organization. I think ultimately, once they've made that decision, then we'll see, I think, us moving higher along that range.
Manav, I think there may have been a third part of your question.
Yeah. There was a second part, which was sorry, Manav, which was the international and what precludes. But I think your question gets it sort of compares two things, Manav. Correct me if I'm wrong. One is you look at our product, the investments in product innovation and the acquisitions. And I hope you can sense our excitement about that. And then you look at the growth rates, and they're in line with the market TAM growth rates, right? And so you're sort of saying, "Well, why aren't we assuming taking more share over time?" Fair question. I will say, and I think Mike will more than agree with this, I would prefer to have the balance be the way we've presented it today than the other way round. I wouldn't want you to be asking the opposite question, which is, "Hold on.
There's not that much in the product pipeline, and yet you're really out over your skis in terms of getting share." We want to be on the conservative side of that as we talked to you today. And we will deliver over the next few years.
Matt?
Yeah. On international, a lot of what Laura just talked about in terms of ONESOURCE absolutely applies for the international markets as well. I think where we have strong content assets or localized assets in our big bet markets that I mentioned, there we already have the content or the necessary presence. And what I think we can see for international growth is applying some of the skills and the technology tools that we have and our global solutions to other markets as well. So I think, actually, it's not really about what's limiting. It's more about where we decide to play, where we choose to be focused to maximize the opportunity in front of us.
Andrew down the front.
Hi. It's Andrew Steinman, JP Morgan. Steve, a question about law firms. Their core business is hourly billings in the timeframe of, let's call it, 2025, 2026. Do you think their business model is going to have to greatly change because of adoption of Thomson Reuters productivity gains products? And if it does, does that affect Westlaw revenues upwards or downwards? And of course, I know Westlaw doesn't charge on a per-seat basis.
Yeah. And appreciate it if Raghu or Emily or anyone would like to jump in on this. This is a huge topic of debate, which is, is the partnership model the right one? Is the per-hour billing going to survive? What does the apprenticeship model look like for young lawyers? What does law school look like going forward? All of these things, if you spend time in large, medium, or even smaller firms, they are hot topics. The first point I'd make is that there's a huge difference between the sort of dependence or reliance on the per-hour billing in the United States relative to the other markets we serve.
So the other markets have, in large part, moved to not entirely, but in large part, moved to more sort of success-based or other models. I personally think we're going to see more of that in the U.S. because I think what we're sensing from Laura's general counsel customers is a bunch of them are going to demand it, not all of them, but a bunch of them are going to demand that, and the firms will respond. And they're going to have the productivity tools from us and others to help them meet that demand.
And that's the biggest difference, I think, going forward, is that they're going to have the tools that enable them to maintain or even increase per-partner profit under a different revenue model. The thing we don't yet know, Andrew, is the pace of that change. My bet is it's going to change. And we're not dependent entirely on that, but it would be helpful to us. The question is how fast the change, particularly in the United States, particularly amongst the sort of the bigger Wall Street firms. But I don't know whether any of our Emily, please jump in.
Yeah. No, I largely agree with Steve in terms of this has been pressure on the law firms for a long time. And when it will happen, I think, is still an open question. The one thing I will say is that for our big legal online products, we've been building efficiency and productivity into those product lines now for years. If you think about with Westlaw Edge, that was about more efficient research, precision, more efficient research. Practical Law has always been sold on the commercial value of doing more with less. So even though we're still seeing this question of when is the billing model going to change, I think we've seen real success with our legal online products in selling against the demand for efficiency from the corporate customers.
Yeah. And I would like to add something to that, right? So one of the interesting trends in the last few years is that the larger law firms have been losing market share to the medium-sized law firms, right, which, again, communicates the sense of pressure from the corporate counsel on the rates. And this is why a lot of the conversations I'm starting to have with customers is, "Hey, it's not about my billing model and protecting my billing model. It's about protecting my business because if I don't take care of this, my business is not in all practice areas, not in all types of matters, but it's going down," right? So I think that's another important driver for this change that I'm sensing from customers.
You still didn't say Raghu. Do you worry that this might affect Thomson's revenues, Westlaw's revenues, or do you feel like this is a law firm challenge?
Yeah. Go ahead.
So again, right, again, with all modesty, I'm coming into this. And my view is we are partners to this industry. And anything that's their problem, it's our problem because they are facing, I think you can argue in two years, three years, five years, a moment of transformation on what this industry represents. And we have both the opportunity and the trust. And we need to be taking their problems as our problems.
That makes sense. Thank you very much. Sorry, Steve.
Good. Scott?
Thanks. Scott Fletcher, CIBC. I have a couple of questions on the cost side. The first is more of a clarification. You mentioned a few times the potential savings from internal use of generative AI. Is that baked into the 2025 and 2026 margin expansion, or is there further upside from the internal savings? It's early stages on that, Scott. As we continue to evolve, that could result into some upside for us. But then the question comes into play, investment opportunities to further accelerate growth versus margin-free cash flow expansion.
So right now, very early stages. And Kirsty will supplement me in a second. I think it'll be sometime in the second semester before we really have good insights onto the timing, sequencing thereof of the internal application of GenAI. But that could result into some opportunities for us. And then with that opportunity, as you get into the investment discussion versus margin expansion. But Kirsty?
Yeah. I think that's actually pretty well said, Mike. It's early days. I think we're super excited by what we're seeing, but we haven't yet set ourselves any hard targets. And then, Mike, as you said, as we get there, we'll need to decide what's the priority, right, just based on where we are in the curve. And I think you can kind of see from David's roadmap, right, we're sort of pretty proud of the team in terms of how much stuff they're managing to sort of get done on the product roadmap. But there's just more ideas and opportunities right now than we can get through.
Scott, did you have a second question, Scott?
I did, yeah. I just didn't think I could speak loud enough. It was just more so on understanding your being driving adoption and trying not to be too aggressive on the usage pricing side of things. If there was a scenario where the cost sort of from the GenAI does ramp up faster than you expect, I guess, on the data call side, how fast do you think you could react, I guess, on the top line to offset any accelerated costs?
Yeah. It's a topic we talk about a lot. Kirsty, do you want to start on that one?
I mean, just while Kirsty's getting a microphone, as you can imagine, Scott, we've modelled every possibility. So we do not see that as a problem at the moment. The way we've been able to price, the way Kirsty's been able to keep our cost structure on this one platform under control gives us a great confidence at the moment that it's not going to be an issue. But Kirsty, please.
Yeah. So I think one of the really good things as an engineer is the number of people that are innovating in this space, right? As I mentioned earlier, we already use multiple LLMs. And the general standard's really high, but some are better for some tasks than others. And as David and the product team put together what they want, we'll test out different things. And we have very much built the platform, as I mentioned earlier, to be plug and play. And so we are looking at what's good enough. And just to give you a really good concrete example, we have launched a lot of our early products with GPT-4, right, not a big surprise. Some of our tests on GPT-4 Turbo actually result in a better answer.
If you don't know what Turbo is, it's basically it just shortens the cycle to make a decision, right, to put it in very sort of simple speak. The answers have actually come out better. So spending less time thinking about it and coming up with a better answer, it's maybe a bit counterintuitive. But that's genuinely what we're seeing. That turns out to be cheaper. I think, for me, it's just really working through scenario by scenario, where do we really need the quality, the pace, the opportunity?
As this market becomes more and more competitive, we will see more LLMs launch. I'm really pleased with the way we've set ourselves up because we can very quickly, in a couple of weeks, deploy a new LLM and get testing on it and see what those results look like. It's something we are very focused on but have high confidence in our ability to manage through that.
Scott, I'd just add the finance team led by Mike, and I'm one of the others on it. We're tracking this exceptionally closely. We've got our arms around it.
I think Heather has a question. Gary.
Hi. Thanks again. Heather Balsky at B of A. Mike, I think this is a question for you. With regards to M&A and as well as organic investments, you talked about wanting to deliver margin expansion, but also it seems like top-line acceleration is really important to the company. So as you look out over the next few years, the opportunities that come in front of you, how are you going to prioritise the top-line long-term compared to delivering on your midterm margin targets? And then as a follow-up question with regards to portfolio optimisation, is there still opportunity there? When you talked about, for example, FindLaw delivering flattish growth, where does Print fit into the organization today compared to a number of years ago and just how you're thinking about that broadly?
You start with the second question first, Heather. I think the portfolio optimization work is never complete. I think we have a fiduciary responsibility obligation just to continuously assess. I commend David Wong, Kirsty Roth in that regard. We talked today about dropping the number of products from 170 to 110 over roughly the last three years. Could there be more opportunities to optimize the portfolio? I think the answer is yes, but we don't have any specifics today, Heather, in that regard. I would just say it's part of that continuous improvement or continuous assessment process. But we have to look at it financially, strategically, customer portfolio effect at all. So I think it's never done. We'll continue to work through it and keep you updated as we move forward. In regards to Print, Print is performing very well.
Jennifer Prescott, the President of that business now into her third year, is doing a great job with Print for us. As you know, Heather, that's a significant contributor of free cash flow. It was nearly $200 million of free cash flow in 2023, which allows us it's a very low capital intensity business, which allows us to funnel that investments into higher areas of the business. On your first question, certainly, organic revenue acceleration, sustained acceleration is really important for us.
Given our operating leverage, we're hopeful that we'll be able to manage the margin expectations as the M&A. But it's going to depend on the size of the acquisition target and where it is in its life cycle, etc. It's hard to give you a specific answer in regards on that. But certainly, accelerating our organic revenue growth is a high, high priority for us. With the operating leverage that we have, that affords us opportunities. But we're focused on both.
I think we have time for one more. Kevin, close us out.
Great. Thanks. And again, just a really powerful presentation. I want to talk about the go-to-market strategy because you have a unique product suite in terms of you can serve the enterprise all the way down to mid to down market. And have you given much thought as to because when I think down market, it seems like with the product suite you have, you can almost create a platform solution as opposed to point solution in the enterprise. And again, I think one of the things we're so focused on is the retention optionality in the model. But just any thoughts as to that platform as opposed to point, and particularly given the optionality you're bringing to the market overall?
Yeah. I'll start. And maybe the presidents can add. Kirsty may have a comment on the platform element. It's worth noting today, I mean, Elizabeth's business serves a very large number. I think the number's up to about 80,000 tax and accounting firms. So today, we already serve that sort of long-tail business. The same.
That was a very direct one.
Wolf always reminds me, "Be succinct." And that was a very direct one.
They didn't like my question.
I mean, long story, but in tax and accounting and to a large extent in Raghu's business, we actually already serve a large number of firms. It's really in Laura's where we tend to skew toward the top end of town. The SMB opportunities, we call it, is there for us. And I think one of the questions is, can we build into that organically with the platform you describe, or can we do it through M&A? And that's sort of a constant source of study for us.
I think Kirsty made the point. In terms of our innovation p roduct roadmap, we've got more opportunities and ideas than we do have capacity at the moment. It's not a financial question. It's much more sort of a talent in getting through it. So I think it's not something we've baked in. This broader SMB opportunity is not something we've baked into the presentation or the numbers today. But it is something we study. I don't know whether anyone else wants to start.
No, I just wanted to add just the acquisition that we made of Pagero. I referenced in my presentation the Smart Business Network. When you think about the underlying sort of theme around Pagero, it starts with e-invoicing because of the mandates. But it also could grow to all of those companies, suppliers around that organization that is complying with that mandate. These are smaller, medium businesses that are out there. So we're always sort of thinking about, how could we serve them too that is providing a platform that is global and open, that's trusted and secure to help with those organizations also?
So it's part of a longer-term vision. But we're always thinking about it. Yeah. I would just add the key difference from a couple of years ago is what we've built with the digital platform. So that allows us to really serve SMB-type customers. We've got that now. I think to Steve and Laura's point, there's no reason why we couldn't do it from a tech perspective. It's potential for the future as we decide what we want to focus on.
From the perspective of the legal industry, I really like your question because lots of the large customers we deal with, they have a lot of resources. They are playing with lots of different technologies. But when you go to smaller law firms, they don't have those resources to choose between three different AI platforms. They can't sustain that. They can't have different software pieces. We believe we are in a good position. That's the kind of thinking in terms of providing one platform for AI and also combining software and content together. I think we are well positioned there.
I think with that, it's time to break for lunch. Let me just close by reminding people, the kiosks outside will be up and running for the next 45 minutes. I really encourage you to go spend some time with our folks out there and see how this stuff works. There's lunch outside that you can bring back in here.
Thanks. Just a last word, Gary. On behalf of the management team and also the board members in the room, just want to thank everyone for your time and your attention and your interest in TR. We greatly appreciate it.