Good day, and thank you for standing by. Welcome to the Wolters Kluwer Investor Teaching webcast and conference call. My name is Brika, and I will be your coordinator for today's event. Please note that for the duration of the call, your lines will be on listen only. To ask a question, you have two options: one, press star one on your telephone keypad, and you'll be put in a queue until your line is open; or two, you can submit a question via the text box below the player, which will be read by our host, Meg Geldens. Please be advised that today's call is being recorded. I will now hand you over to your host, Meg Geldens, Vice President, Investor Relations, to begin today's call. Please go ahead.
Thank you, Brika. Hello, everyone. I'm Meg Geldens, Head of Investor Relations at Wolters Kluwer. Thank you for joining this event, which will focus on artificial intelligence. AI has been, by far, the number one topic of discussion with investors and analysts in the past four months. So today, we brought together several members of our team to tell you about our AI technology and two of our largest product lines that are leveraging advanced AI to meet customer needs and use cases. Before I introduce the speakers, can I ask you to read slide two about forward-looking statements? We caution that any forward-looking statements made on today's call are qualified by certain risks and uncertainties that could cause actual results to differ materially from what is contemplated by these statements. We will not be disclosing new information on current trading today. Now, it's my pleasure to introduce today's speakers.
Nancy McKinstry, our CEO, will start us off with a brief introduction. Alex Tyrrell, CTO for Health and leader of our AI Center of Excellence, will talk about the DXG organization, how we develop our platforms in close collaboration with our customers, and how we're building AI into our solutions to drive enhanced value and enhanced decision-making. Next, Jason Marx, CEO of Tax and Accounting, will put the spotlight on one of the fastest-growing solutions in our North American tax and accounting business, the CCH Axcess Cloud Platform, which is designed for professional accounting firms. Then Cathy Rowe, head of the professional segments within our US tax and accounting group, will demonstrate one of the six agentic AI tools that we are rolling out across CCH Axcess.
After that, Greg Samios, CEO of Health, will provide a deep dive on UpToDate clinical decision support, its unrivaled body of content, and how its new conversational AI interface delivers on the needs of our enterprise hospital and other healthcare Julie Frey, VP, Product, will then give you a brief demo of the UpToDate Expert AI. Finally, Stacey Caywood, member of the executive board who will succeed Nancy as CEO next year, will summarize her strategic priorities, how we expect to make a return on our AI investments, and why we are well-positioned for growth in the future. With that, I hand it over to you, Nancy.
First, I will take you through how the tax professional sets up their work for the purpose.
Start off by reminding all of us how, over the many years, we've transformed this company in so many ways. Through organic investment and portfolio actions, we have transformed from a fragmented, print-centric holding company into a digital expert solution company, which is now being powered by Expert AI. Approaching 70% of our digital revenues come from products that are AI-enabled. This transformation has accelerated our organic growth and increased our margins as we continuously enhance the value we bring our customers: innovations that improve their analysis, their decision-making, and their productivity. Perhaps less visible to all of you on the outside is the transformation in our operations and in technology. Over the years, we have centralized all of our core functions, from product development to HR to finance, among others, in order to drive excellence and scale in these areas.
Our transformation and our focus on customer-led innovation, coupled with substantial investment, has brought us to a place where today our portfolio is stronger than ever. We have the key assets, market positions, technology platforms, and talent to take advantage of the latest round of next-generation AI technology, particularly agentic technology. It is part of our culture and our strategy to keep innovating and transforming. We don't stand still. We are continuously reinforcing our competitive strengths, whether that is our data, our customer relationships, our technology, or our brands. In product technology alone, we are investing some 7, sorry, 650 million euros this year. Much of that is going towards product development powered by AI. AI offers Wolters Kluwer substantial opportunities to leverage our proprietary data and content, our leading market positions, our deep expertise, and our advanced technology architecture to drive faster growth and deliver enhanced value to our customers.
Today, we will give you some insights on how we have centralized our technology team and what benefits that brings. Before we do that, it's important to know that over many years, we have been streamlining, standardizing, and advancing our technology. This continuous investment puts us in a strong position to develop AI-powered platforms in a fast and efficient way. In our market segments, we were among the first to invest in cloud platforms and to build APIs that connect into ecosystems. Our early move to cloud-native architecture is giving us an advantage as it enables more seamless AI integration and delivers a better user experience. We have also been on the front foot in embedding AI. Starting some 12 years ago, the proportion of our digital revenues that is AI-enabled is approaching 70%, which is up from 50% just two years ago.
Every month, we are rolling out new AI tools and enhancements. More recently, the technology team has created a proprietary AI-enabled platform called FAB, which is allowing us to speed up development cycles and drive economies of scale. You will hear more about that from Alex shortly. Our strategy has remained consistent throughout. We combine trusted expert content with advanced technology to bring increasing productivity and improved outcomes to our customers. This strategy has delivered sustainable and profitable growth and positions us well for the future. With that, I'd now like to hand it over to Alex to talk about DXG.
Wonderful. Thanks, Nancy, and hello, everyone. I'm here to give an overview of our technology organization and AI strategy. Let me start by describing our current product technology organization. The Digital Experience Group is a central technology organization that supports all five divisions. Our mission is to drive faster customer-facing innovation and to leverage our technology investments to drive efficiency and scale. We are vertically aligned with a CTO for each division, ensuring our technology strategy stays aligned to business needs. We offer reusable and scalable assets to all five divisions of Wolters Kluwer, with approaching 6,500 technologists in support. We also maintain a number of centers of excellence. They are aligned horizontally, which provides uniform access to specialized talent, enhances our ability to scale while ensuring proper governance.
Importantly, the AI Center of Excellence is a key part of our innovation strategy, really focusing on how GenAI can transform our products and services and deliver the right outcomes for our customers. We've been investing in AI for about 15 years. As part of our strategy, we established our AI Center of Excellence 12 years ago. Now, from day one, our strategy has been consistent: use AI to deliver measurable end-to-end outcomes for our customers' most important workflows. We have continuously supported this strategy by investing in the technologies and platforms that power workflows at scale. At each point in our AI journey, safety and trust have been hallmarks, and we always have grounded AI in our curated and verified content. Today's AI technology lets us accelerate this strategy by building on our prior investments and continuing to widen our competitive advantage.
Now, on the left, you have traditional AI and machine learning, which has been sort of our historical focus: predictive models on structured data, often rules-driven and narrow, good at specific tasks like document review, but humans still had to stitch the workflow together. Now, in the middle, we have GenAI, where large language models have represented a step change in AI, unlocking content generation, things like text, images, code, but mostly a single-turn transactional moment, like drafting an email or generating a contract clause. Productivity goes up, but the loop isn't closed. On the right, agentic AI. This is where the technology really accelerates our strategy and builds on our prior investments. Agents plan. They take actions and interact with enterprise systems, augmenting and assisting users by executing multi-step workflows on their behalf.
Importantly, we have made continuous investments in our cloud solutions, our APIs, and platforms that are the necessary prerequisites to make agentic AI work at scale, to solve hard problems and deliver the right outcomes for our customers. Now, in terms of value creation, our decades-long history curating trusted content to drive our AI, as well as our investments in scalable technology assets built to support workflows, gives us a unique competitive advantage. Agentic AI builds on these prior investments, giving us the foundations to capitalize quickly, continuing our delivery of durable, high-quality growth for shareholders. Now, to compete and win in this new agentic world, you need a deep understanding of the customer and how their work actually gets done. When you truly understand the workflow, the steps, the frustrations, the information gaps, you can streamline or even automate a complicated end-to-end process.
Tasks that once had to be done sequentially because information was scattered or systems didn't connect and required different people to manage can now be executed in smarter and more efficient ways, taking advantage of parallelism and increased automation that is orchestrated towards the right outcome. The next step in our evolution is workflow transformation using agentic AI. A key to this transformation lies in our subject matter experts, who know the jobs to be done and can translate that knowledge into the design of agentic systems. Their collective skills and expertise curating the trusted information that our customers have relied on for decades is now unlocking new value through agentic AI, and this is where our Expert AI strategy creates meaningful competitive advantage. Our approach starts with something no general-purpose AI can replicate: deep domain expertise combined with our trusted and verified content.
Our specialists understand exactly how professionals work, the decisions they need to make, the information they trust, and the risks that they must manage. We capture the experience and knowledge of our specialists, translate it into structured guidance, and adapt AI models to think and act like experts, all grounded in our proprietary content, preventing hallucinations and establishing trust. We also fine-tune and capture expert reasoning through techniques like chain of thought, all reviewed and continuously improved by our domain experts in the loop, working closely with our AI engineers. For complex tasks, we may adapt multiple models, each capturing a unique aspect of how experts perform work. We will see an example of this later when we demo UpToDate Expert AI.
Now, this creates an AI capabilities competitors cannot easily copy because it's built on our decades of proprietary expertise and content that is guided at every step by experts in the loop. Now, Expert AI is a cornerstone of our strategy, which we need to execute quickly at scale without compromising on safety. This is made possible through our proprietary AI enablement platform that we call FAB. Now, the vision for FAB is to move beyond foundation and frontier models to solve end-to-end workflow. This unlocks new value in our decades-long investments in scalable technology assets and proprietary content, which increases our competitive advantage. Now, FAB development begins with high-value use cases and clear ROI targets. As part of our governance framework, we immediately establish guardrails at the earliest design phase, ensuring responsible and ethical use of AI.
Within the FAB ecosystem, we provide a number of key features and innovations that drive scale and adaptability. Foremost, we focus on model pluralism: pick the right model for the right job and adapt quickly. From there, we provide the ability to fine-tune and ground models using our trusted content. In the middle is our agentic layer, where we build reusable capabilities, allowing us to scale. Now, Pinboard and Labs provide low-friction environments for orchestrating agents and running le an experiments. In XBit, we have our standard UX design patterns, ensuring our interfaces are consistent and intuitive. And finally, Bridge is our gateway to external systems, extending the reach of FAB into new ecosystems while remaining safe and well-governed. Now, undergirding our FAB ecosystem is a suite of proven GenAI productivity tools that accelerate delivery. These tools target the entire development lifecycle, from design to writing code, testing, and deployment.
As we will see in today's demos, FAB is delivering significant value for our customers today. With FAB, we are AI-ready in the enterprise on day one: professional-grade, trusted, and secure, not bolt-on, not point solutions, which is key to scaling our Expert AI strategy, and with that, I will hand over next to Jason Marx.
Thanks, Alex. My goal today is to provide a clear understanding of how Wolters Kluwer applies the AI and agentic AI workflows in tax and accounting, what it looks like in practice, and why that matters for tax and accounting professionals. Tax and accounting is one of Wolters Kluwer's growth engines, a leader in the professional accounting firm market, delivering $1.6 billion in revenue last year, with approximately 60% of those revenues from North America. As a global provider, we serve tax and accounting professionals with cloud-based AI-enabled solutions that help them manage compliance, optimize workflow complexity, and deliver value to their clients. The flagship platform is CCH Axcess, a cloud-native suite of 16 integrated modules that embeds Expert AI across tax, audit, and practice management workflows, and the scale is significant: 10,000 firms, including 95 of the top 100, and 1.4 million users of CCH Axcess.
And this scale and this breadth of a platform is a critical differentiator. CCH Axcess is the only true digital core in tax and accounting: a unified cloud-native platform where AI is built in, not bolted on. And this is the foundation that firms need for the agentic future: to move from generative to embedded to agentic AI with unified data, expert embedded content, rigorous security, audit trails, and explainable outputs. In high-stakes professions with regulated compliance, expert judgment and nuance are critical for accounting professionals to trust the solutions that they use. With CCH Axcess, this is made possible through decades of proprietary expert-validated content, the compliance foundation of CCH Axcess. To do this, we analyze over 180 primary content sources. We process 250,000-300,000 regulatory changes annually across federal, state, and local domains. And at the heart of CCH Axcess is a layered intelligence model.
Now, it starts with public tax and regulatory data, which includes U.S. federal tax legislation, state codes and regulations, and even local codes and regulations. Then our editorial experts, they analyze this information. They enrich it with proprietary content, including CCH Tax Analysts, expert insights, published content from CCH editors, and detailed state-level analysis through our CCH State Tax Reporters. Then we embed the compliance and calculations across federal and state tax forms, and we integrate both firm-specific and client-specific data that would come from sources like the prior year tax return. All of this comes together inside CCH Axcess, creating a single intelligent platform that combines public data, proprietary expertise, and client context.
Our Expert AI amplifies this foundation, delivering trustworthy expertise with capabilities that drive efficiencies for the firm: smart document collection that predicts needed information and streamlines client uploads, intelligent processing that automates classification and routing with expert oversight, and proactive insights that are able to surface advisory opportunities by the CPA before their clients even ask. And the result? Professionals spend 20%-30% less time on manual tasks and more time delivering strategic guidance and advisory services. The CCH Axcess suite is a cloud-native platform of 16 integrated modules, unifying tax, audit, and firm data into an intelligent, scalable experience. Now, this past October, we announced the commercial launch of six new AI-powered modules that you can see here around the core.
And because it's a unified platform with Expert AI embedded throughout, we can deliver an agentic-ready experience, one where we can orchestrate workflows and automation across the modules with trusted compliance built in. Take, for example, a typical tax workflow. Now, the ultimate goal here is a no-touch tax return. And with Expert AI agents and APIs, we can extract, classify, and import data directly into the return. Our Expert AI agents coordinate tasks and externally import client financial data, things like bank statements or accounting data, and they do it securely and at scale. And this is the foundation to move customers to full agentic task orchestration. And the entire process on CCH Axcess will evolve to be managed by agents, from initial data gathering and Client Collaboration to initial return preparation delivered to the CPA for final review.
Then agents coordinate e-signatures and e-filing, all within a single integrated, secure digital environment. For the accounting firm, this drives efficiencies and frees up time for higher-valued advisory work. And for WK, this drives opportunities for upsell across modules and greater stickiness and retention of firms across the CCH Axcess platform. Cathy Rowe, our Senior VP for our professional business in the U.S., is going to do a deeper dive and highlight and demo the power of CCH Axcess Client Collaboration. It's a great example of how AI moves from concept to practical application. So let me set the stage for what you're going to see. CCH Client Collaboration transforms the firm-client engagement by automating the intake, the document exchange, and the communication between the CPA and their client. Why does this matter? Because these steps are often the most time-consuming and error-prone in the tax preparation process.
With Expert AI and unified data, interactions become predictive and proactive, moving us one step closer to the no-touch tax return. Here, agentic AI is contextually aware of the unique circumstances of each taxpayer, anticipating what information is needed, reducing the back-and-forth nature of the process between the CPA and the client. And the benefits? Well, they're clear and immediate: time savings. They free up staff for higher-value work. We elevate the client experience, which for the firm builds trust and retention. Operational excellence accelerates responsiveness and improves accuracy. And importantly, these efficiencies create capacity for the firms that are able to grow and offer new, highly valued advisory services. This is a great example of how the power of Expert AI becomes both trusted and tangible for the firm. Now, I'll turn it over to Cathy for the demo.
Thanks, Jason. In just a moment, I'm going to show you an end-to-end integrated tax experience that will leverage Expert AI to achieve touchless, ready-to-review workflows. This starts with how the tax professional interacts with the taxpayer. The goal is to have a frictionless experience from signing an engagement letter all the way through obtaining, understanding, and populating that and finalizing the return. As Jason noted, AI is leveraged throughout the entire process to help understand and harness the data to get insights and automation. In the next few minutes, I will take you through this experience that enables review-ready tax returns, leveraging agentic AI to set apart the client experience for the firm. First, I will take you through how the tax professional sets up their work for the purpose of engaging the taxpayer.
By understanding what was done for that taxpayer in a prior year, we can create a request list effectively by using work previously completed. We also give customers the ability to automatically generate an engagement letter, essentially a contract for the work. These will both be created with AI, keeping expert in the loop. Next, we will look at the taxpayer experience. The goal is to make it as effortless as possible for the taxpayer to provide what the tax professional needs. After a simple login and seeing what's changed, they can provide the documents needed with just a simple drag-and-drop everything at once. Imagine bringing a shoebox of documents with all kinds of data into your tax professional. This is exactly what the taxpayer is doing here.
AI automatically identifies, classifies documents against the request list, showing what's provided, what's missing, and whether enough was supplied for the tax professional to proceed. The CPA immediately sees these updates and then can see the phase of the tax engagement. Subsequently, when the return is prepared, the taxpayer reenters the experience, sees whether they are due money, whether they owe money, can download their return, and ask questions of their data, review and sign the necessary forms, and then get their CPA to file. Once the return has been completed, there is another Expert AI element for the taxpayer to engage. They are shown their completed federal and state returns, enabling the taxpayer to ask questions about the return, again using Expert AI capabilities. Now, switching to the tax professional's perspective, the tax professional can see what's happening across all the tax returns for the firm.
AI is used to summarize the activity, and there is a list management area that shows statuses that will update automatically. Expert AI highlights what needs attention, whether there's follow-up tasks, and accordingly, the system will update automatically. The data that was previously provided by the client is made available directly here to the CPA, so they are able to review and then flow directly into the return, at which point we have a touchless, ready-to-review tax return. Now, I will hand it over to Greg.
Thank you, Cathy. Hello, everybody. I'm Greg Samios, CEO of Wolters Kluwer Health, and welcome, everybody. My goal today is to provide you with a deeper understanding of UpToDate, its proprietary content that's designed for point of care, its integration into the healthcare ecosystem, and the Expert AI interface that we are busy rolling out right now. So let's get started. UpToDate is the largest single product in our health division, and it's part of our Clinical Solutions group. It offers a range of point-of-care solutions for hospitals, pharmacies, payers, and other providers in the healthcare ecosystem. UpToDate is the global leader in clinical decision support, trusted by more than 3 million clinicians who consult the application 50-60 million times per month. It consistently has a Net Promoter Score above 70 and retention rates above 90%.
About 80% of UpToDate revenue comes from enterprises purchased by the C-suite and buying committees of hospitals and other institutions. The remainder comes from individual clinicians and small practices. Uniquely, UpToDate's content draws on the knowledge and experience of our more than 7,600 clinical experts who are among the world's leading practitioners in their area of medical practice. Over the years, UpToDate has evolved from a single product that provides decision support for clinicians into an integrated platform that drives clinical outcomes for the enterprise. This enterprise platform touches every part of the care journey. Step inside a hospital and you will see it immediately. Physicians rely on our clinical decision support at the point of care. Nurses rely on our drug data to administer drugs. And when the care team discharges patients, they send them home with our educational materials.
Administrators monitor and improve quality metrics through our analytics and dashboards. We pair that with deep workflow integration, API connectivity with all the major EMRs, and partnerships with the leading ambient scribe vendors, ensuring that UpToDate Enterprise is embedded directly into the clinician workflow. So whether a doctor is looking at a patient chart in Epic or using Abridge to capture the encounter, the power of UpToDate is at their fingertips. This is the power of a common, trusted evidence base across the enterprise workflow. It has taken significant content harmonization and technology effort to get here, and we continue to expand the platform. Most recently, we've introduced UpToDate Expert AI, giving clinicians fast, accurate answers to their questions grounded in our proprietary content.
But before I discuss our new AI initiative, let me describe our unique editorial model that ensures we always have the latest clinically relevant evidence. We follow a rigorous editorial process that's been refined over 30 years, which results in what is considered the gold standard in clinical decision support. More than 7,600 leading experts, supported by our 100-plus in-house physicians, nurses, and pharmacists, continually evaluate new evidence and translate that into actionable point-of-care guidance. Together, we monitor thousands of primary sources, including medical journals and clinical trials. But interestingly, only a small portion of primary content is relevant for clinical practice. And this is where the experts are critical. They're able to filter out research papers that are not clinically relevant, as well as those studies that might be statistically interesting but don't change clinical practice.
They also apply their clinical judgment when new medical evidence falls into the gray zone, which is quite often. Only about 20% of recommendations in UpToDate are Grade 1A, meaning supported by overwhelming evidence where there is a clear yes or no answer. The majority of recommendations are underpinned by our experts who apply their judgment to identify exceptions, contraindications, population-specific considerations, as an example. These aspects cannot be derived from research journals alone, and finally, our experts explain what the findings mean and how to apply them. Think of diagnostic pathways or algorithms or dosing considerations or workflow implications. Our integration partnerships with EMRs and ambient scribes and others ensure this trusted content reaches clinicians at the moment of decision. Increasingly, we're pairing patient data or ambient clinical conversations with UpToDate content to deliver tailored, real-time decision support.
And we do this through a bring-your-own-license model that protects our IP. And importantly, our process is fast. Practice-changing updates often appear in UpToDate well before the research is published or before the society guidelines are updated. This has real-world impact. Roughly 30% of the time a clinician uses UpToDate, they change their treatment plan, driving better outcomes and well-documented ROI for health systems. And that's important. This slide shows two examples of practice-changing updates that we made in 2025. You can see the details, but to summarize, the first example we show how our oncology experts were tracking emerging evidence very early. And when an oral abstract at the ASCO Symposium showed promising randomized trial results, our experts convened, debated, and ultimately issued new recommendations before the publication in NEJM and before the NCCN guideline revision.
In the second example, an NEJM study on methotrexate as a steroid-sparing option showed promise but wasn't definitive. Our pulmonary specialists debated which patient groups would benefit and where steroid therapy remained essential. The result was a nuanced practice-changing update in UpToDate, where society guidelines remained unchanged. Therefore, a clinician following only published research or guidelines or using an AI tool that relies on them could end up with a suboptimal treatment plan. In addition to their role in maintaining our gold standard content, our experts are now playing a critical role in developing our Expert AI technology. In high-stakes clinical settings, even an occasional hallucination is unacceptable, and to address this, our Expert AI technology uses multiple large language models, more than 30 proprietary AI agents, and uses expert-in-the-loop in two AI-powered layers on top of our foundation content.
The combination of technology and expert oversight is what makes our clinical reasoning unique. Our patent-pending clinical reasoning layer breaks the natural language question into core elements, assembles the decision pathways, identifies what proprietary content is needed, and then evaluates relevance and produces a synthesized answer. Next, our validation layer, which is based on expert-developed rubrics with guardrails and answer sets, scores the quality of the AI response, ensuring accuracy and tightening the model over time. UpToDate Expert AI is purpose-built to accelerate time to answer, but importantly, to deliver the accuracy, clinical relevance, and workflow integration that our customers need. We are relentlessly focused on delivering the right information at the right time to the right person in the right modality in the right format. In discussions with customers, our triple-layer expert-in-the-loop process is a meaningful competitive differentiator.
The foundation layer, the clinical reasoning model, and the validation process that leverage our experts are so important in high-stakes decision-making. Drug dosing is a good example. Highly complex, high-risk drug events are the single largest category of adverse events in the hospital. Simple AI tools lack the full patient and clinical context required to get dosing right. The UpToDate suite includes comprehensive drug dosage and indications fully harmonized with the UpToDate clinical content. Another differentiator is our AI answers, which are optimized for point-of-care use. Not only are they accurate, but they're accompanied by clear assumptions and treatment considerations that prevent wrong conclusions and missteps, and by clinical nudges that support a richer diagnosis. Integration into clinical workflow is what truly differentiates us. We are deeply embedded at the enterprise level, integrated into major EMR vendors, and expanding through partnerships and ambient scribes.
I'm really pleased to see that we're making rapid progress in the rollout of UpToDate Expert AI. Two years ago, when we first showed an AI interface to hospital customers, many had reservations about using AI in their high-stakes decision-making. Today, after extensive collaboration, we are in full launch mode with a product that meets their needs and addresses their concerns, and customers are ready to adopt. In just 45 days, we've had over 80 enterprises sign up to take Expert AI. So far, we've gone live with over 20 enterprises. All have completed a rigorous governance and security review of our new product. Among the 80 preparing to go live, our enterprises in key international markets where strict privacy laws will be a challenge for any ad-based tool. We've launched UpToDate Pro Plus for individual physicians.
It's early days, but currently, about one-third of doctors up for renewal are upgrading to the AI package, highlighting the value we deliver, and the pace of innovation continues. We've delivered eight new releases in the last 45 days, pushing the bar on capability, performance, speed, and drug coverage. As a team, we are really excited about the launch, how the launch is going, and with the impact that it's having on our customers and the patients they serve, and now, I'll hand it over Julie Frey for a short demonstration of UpToDate Expert AI. Julie?
Thanks, Greg. Hello, everyone. I'm Julie, VP of Product at UpToDate. As we've been bringing our AI capabilities to customers, the feedback has been truly energizing for our team that's hard at work. Customers tell us they're especially excited about giving trainees, medical students, residents a trusted, approved AI tool right in their workflow.
What you're about to watch is a demonstration of UpToDate Expert AI. So let's go ahead with the demo. Let's look at a case that looks simple but is potentially high risk. So we have a pregnant patient, 36 weeks, whose water has broken, but who's not yet in labor. Is it safe to send this patient home for a few hours of observation? This is a scenario where accuracy is non-negotiable because the risks - infection, hemorrhage, cord prolapse - they can escalate really, really quickly. So let me take you through how UpToDate Expert AI handles this question. What I'll point to you first is our assumptions. They're a really unique part of our offering.
But the way that assumptions work, the intent behind it is to mirror the way a clinician thinks, as well as to prompt the user to ensure that they've added everything that is relevant to answer the question. That is one of the first areas that we found our users really focus on and really love about the solution. We also always, above that, will note that this is, we're using AI as part of the solution. This is critical to be transparent. We actually have taken transparency really to the next level. We show users every single step that we take in order to respond to this query. I don't expect that clinicians will always have the time to go into all of that, but it's available to them, and should they want to use it, they can access it.
The assumptions, though, is really the first unique pillar of what our offering and our solution is. Then let's go into the second pillar, which is providing clear evidence-based management guidance that is grounded in UpToDate's expert-curated content. It really tells you directly that hospitalization is recommended. That's in alignment with ACOG guidance. It doesn't stop there. It includes treatment considerations. That's really the second pillar of differentiation. It explains the risks of infection and neonatal sepsis. It outlines why induction is preferred, and it gives the criteria that would need to be met before even considering home monitoring. These are considerations that encourage critical thinking. They reinforce that home management applies only in extremely select cases. They point out that the risks a clinician may face as they're making these decisions and that they really need to be aware of.
And then finally, they summarize the key safety issues in a way that guides the next safe action. Below the response, you'll see a section titled "Learn More." This is actually what we internally call our clinical nudges, another unique component of our solution that our users so far have really, really provided a lot of positive feedback on. What we're focused on here in this section is encouraging the clinician to remain curious and to think of the optimal next best set of actions, even if those are not intentionally stated in the question. The clinician and the clinician's autonomy and agency to take care of the patient is paramount to the way that we structure this. To summarize, by including our assumptions, our treatment guidance, and then ultimately our clinical nudges, the output is truly optimized for the point-of care use case and will drive optimal clinical outcomes.
We tested the same scenario across several leading other LLMs, medical and otherwise, and some did answer the question correctly. But even when they were answering them correctly, they gave such a definitive tone with such confident statements, such an approach that you risk not encouraging the clinician to remain curious and be at the center of the decision-making. In others, in fact, we found that they recommended home management for 12 hours is reasonable. In other words, not immediately going into a hospital or a clinic to be monitored. Some even cited home management for 12 hours was supported by leading medical societies. This is not correct and can be dangerous. And because you have this very definitive, confident tone, the model can unintentionally teach the incorrect practice and ultimately could lead to deskilling of our clinicians. And this is really the core clinical challenge that we're faced.
How does a clinician benefit from the speed and the information processing power of AI without endangering the patients? Finally, I'd like to point you to a unique component of our solution. I'll zoom in a bit there so you can see it. We are gathering feedback really at every step of the process. Users can give us a simple thumbs up or thumbs down. We're also encouraging them to give us more information around what they like or don't like. And just as a headline, we have very favorable thumbs up, which we're really encouraged by, and we continue to strive to do better. But most importantly here, on the thumbs down component of this, we have one of these pre-selected options, which is a critical concern. Effectively here, this is the opportunity for a user to flag anything that they're really worried about with this response.
So think of patient safety or some sort of harm concern. If a user selects this and clicks submit, this then goes directly to our clinical team. We have a team of clinicians whose sole responsibility is reviewing feedback, prioritizing that feedback based on patient safety, accuracy, and harm, and triaging that. We have a 24-hour response time for any of those critical concerns. And then we will get back to the user and we'll have a dialogue with that user around their concerns. It's really important to have that feedback loop. If there is anything around our clinical intelligence, our content, our validation framework, our safety framework that needs to be adjusted to improve, we'll take that as a priority and act on that very quickly.
We really see this as this level of transparency and this level of working with our users to continue to enhance the system and optimize it for the point of care. The transparency that we want to provide behind this is really paramount to the way that we're approaching, uniquely approaching, delivering an AI and our enhanced solution tuned for the point of care. I'm especially heartened to see users giving us an 80% plus score on that thumbs up feedback mechanism that you just saw. With that overview of UpToDate Expert AI, I'll now hand things over to Stacey.
Thank you, Julie, for that great demo. And let me add my welcome and thanks to you all for joining us today. Our expert solutions support professionals in their mission-critical work.
They help them make high-stakes decisions and stay productive at a time when many of our customers are facing real talent shortages. As Jason and Greg highlighted, these expert solutions are now evolving into AI-powered modular platforms. Our proprietary data and content, along with our deep understanding of customer workflows, allow us to identify the right use cases and extend our value proposition. Our foundational AI capabilities, including FAB, which Alex highlighted earlier, combined with our significant cloud investments, are helping us build and scale these agentic use cases efficiently across the portfolio. As we evolve into broader platforms, we orchestrate more complex workflows that draw from multiple content sources, including data held by our customers. Our long-standing experience with APIs and secure data exchange allows us to connect seamlessly into customer ecosystems, whether directly or through partners. For our customers, this delivers meaningful productivity gains and better outcomes.
For us, they drive stickiness and open clear monetization pathways. Our approach to monetizing AI depends on the value delivered and the market context. Historically, we monetize continuous innovation through regular value-based subscription price increases. This approach has long supported our organic growth and consistently delivered strong returns on our investments. Expert AI with UpToDate Enterprise is a great example of this. In some use cases and markets, we're now seeing a clear willingness to pay more for specific AI features. This is especially true where customers see tangible productivity gains or cost savings. And this can take the form of a premium package or add-on modules. As an example, UpToDate Pro Plus for individuals includes Expert AI and other value-added features for a 20% uplift in price above the standard version.
As Jason mentioned, the CCH Axcess platform with its 16 modules, including several AI-powered ones, is a great proof point where customers purchase additional modules to leverage productivity gains from our AI solutions. We expect AI pricing models to continue to evolve. For certain use cases, consumption-based pricing may become a standard component, especially where a customer's AI usage drives variable model run costs. As you can see, we tailor our approach to use case and product context and see multiple pathways to upsell as we build out our platforms. I am proud to say we are moving from strength to strength, and today's event has demonstrated that we have the key assets, including talent, proprietary content and know-how, and trusted customer relationships to win in an AI world.
We start with a strong foundation, a business that has over many years continuously deployed the latest technology combined with our deep domain expertise to deliver value for customers and reinforce our competitive advantages. When it comes to our strategy, I'm laser-focused on a few areas. One is to accelerate the pace of innovation, and this is well underway. Our FAB AI enablement platform has already helped speed up our development cycles in 2025. We will fund our AI investments while still improving margin by shifting the focus of our roadmaps to agentic development and increasing our capacity through the internal deployment of AI tools. A second near-term priority is continuing to develop strategic partnerships. We strive to be fully embedded in our customers' workflow and ecosystems. Partnerships will be important in some markets to deliver end-to-end solutions. A third area of focus is enhancing our go-to-market capabilities.
I'm making sure we are supporting sales teams with data, technology, and the insights they need to capture revenue opportunities. We all know AI is moving fast and will fundamentally change how professionals work. With the strength of our AI roadmaps and the early adoption and strong feedback we're receiving from customers, the team and I are confident that our GenAI and agentic solutions will be a driver to expand value for our customers and propel growth for Wolters Kluwer. Now, I know you have questions, so Operator, if you could please move to Q&A.
Thank you, Stacey. We will now move to questions. To ask a question, you have two options. Press a star one on your telephone keypad and you'll be put into the queue until your line is open.
Or submit a question by the text box below the player, which will be read by our host, Meg Geldens. And just as a reminder for the phone line questions, please press star one on your telephone keypad. Otherwise, you can submit a question by the text box below the player. The first question we have comes from Nick Dempsey with Barclays. Your line is open.
Yeah, good morning, good afternoon, guys. I've got three questions, please. So the first one, there was a recent study, I think it was posted last week to arXiv, showing that for some medical questions, particularly USMLE style questions, generalist AI models were scoring higher for accuracy than the specialist ones, Open Evidence and UpToDate AI. That's what was tested. Can you maybe talk about how relevant those types of medical questions are for the world of point-of-care decision-making?
Is it concerning that those generalist models scored a bit higher in that one study? My second question, DynaMed had DynaMed AI out there, I guess, a year ago or a year, sorry, before UpToDate AI. How was it that they were able to get an AI offering out there that much earlier with a fairly similar type of offering? And my third question, what is Harvey doing in tax that you can see? Can they be a threat to some of your plans on workflow? Is that something that's on your competitive horizon?
Yeah, just to interject here for that first question, I think it would be useful if Julie, Alex, and Greg comment on this. Julie, I think you're very close to our own testing. So if you could perhaps kick off with that one, and then I will come back with the other two
. Sure, Meg.
Happy to take a first pass of this. I think we're going to see lots and lots of different tests come out. I think what I would really start with is we have a number of internal tests that we do consistently. They're all really tuned for the point of care. That's where we're focused. And we see that we consistently outperform both the frontier models as well as ones tuned for medical use cases. And I would also say that we do still consistently see hallucinations and other things that are unacceptable when clinicians need the right answer at the point of care. Alex, anything you wanted to add?
Yeah, I might add, you know, I've read the study. It's interesting.
I think the study even itself alludes to the fact that many of these generalist models have been trained on USMLE and other benchmark test sets, as well as having model alignment performed using the same data set. So there could be some potential for bias there, that they're more focused on the factoid type questions and not necessarily what we look at in terms of clinical decision support. The other finding is that they're introducing potential model knowledge, which, while it could potentially add more context, more clinical context, which could be important, you run the risk of hallucinations. And I think that's a key difference.
Yep. And just to close off, just to remind everybody, we were built for point of care.
Going back to our Value Wheel, we've integrated and spent years harmonizing our diagnosis, treatment, and drug information with our patient education, and being designed for point of care with that precision and accuracy is critical for our solutions.
On the DynaMed AI question, perhaps Greg, you could start off with that one and maybe Alex also on the speed of our development.
Yeah, let's go back to our journey so we started two years ago. We worked very close with our customers in our first AI interface, and at that stage, it was too early for the enterprise level and the standard of UpToDate. Over the last two years, we've worked with those customers to design a clinical-grade, high-quality result, and over those two years, we believe today is the right time and at the right level of quality to deliver to those customers.
So while maybe others would come out earlier, UpToDate has set the bar for the standard for clinical decision support. And so over that time, we worked with our customers to deliver a solution that meets the UpToDate standard. And so that's the driver for our timing.
Yeah, and I might just add that, you know, we co-developed, as Greg mentioned, with our enterprise customers in very high-stakes environments. Built for day one for scale, safety, and an emphasis on privacy and security, making us the first truly enterprise GenAI solution for clinical decision support in the market.
Yep. And maybe one more comment on that. At the enterprise level, as you know, it's very important to demonstrate transparency and trust. And UpToDate Expert AI, which you saw from Julie, demonstrates that. So that was a key addition for our launch.
And then on the question about what is Harvey doing in tax and accounting, perhaps Jason, you can comment on that.
I can take that. Thanks for the question, Nick. So CCH Axcess has strong and a durable competitive advantage that creates barriers to entry for competitors. So we have a deep domain knowledge, close relationships with our customers, proprietary data, and importantly, because it's a cloud-native platform, we're contextually aware of the client circumstances. So we can, as Cathy showed you, we can connect what's happening with an AI, an Expert AI solution like Client Collaboration with the unique circumstances of each individual client that the CPA is working for. And it's only because we have that cloud-native agentic-ready workflow that we're able to do that around the core of the 16 modules with deep integration around that product suite. Great.
Thank you. Operator, can you prompt for questions?
Otherwise, I have two from the web.
Absolutely. Just a reminder, if you would like to ask a question, please press star followed by one on your telephone keypad now. Otherwise, you can type a text question in the box below the player. Your next question on the phone lines comes from George Webb with Morgan Stanley. Please go ahead.
Hi guys, thanks for doing the presentation. I've got a few questions. Firstly, maybe just on the process with regards to enterprise customers on UpToDate onboarding to Expert AI. What's the type of governance checks you see them going through to onboard the solution in terms of those kind of 30 of the top 100? And how do you think about maybe the steps that the other enterprise customers would maybe need to go to? Second question, more of a kind of a midterm roadmap.
When we talk about UpToDate, you know, we're still talking a lot about the content and how you can build out the kind of suite of solutions around the content. But on maybe some more tangential areas like AI scribes, you've chosen to partner rather than build native capability. Do you want to play on the broader clinician workflow, which we're seeing maybe other players try to go after? Or are you taking more of a kind of UpToDate content everywhere type of approach? And then just the final question. On the enterprise edition, you know, we've talked about the AI functionality being part of the annual price increase. How do you see the ability to capture back the potential cost of serving from an LLM perspective, obviously as a cost of doing those queries? Thank you.
So I think quite a few of those are for you, Greg.
So if you could talk about the process that you go through with governance and security checks and then the partnerships with the scribes, etc.
Yes, that'd be great. I'll start it and Julie jump in as well. The governance process, yes, so we've worked very closely with our enterprise customers. What we find is very important, as I mentioned before, the transparency and understanding exactly where the answers are coming from, so black box solutions are very tough for them. If you looked at the demonstration, we showed sort of each step of our reasoning model, and knowing that the information is coming from a trusted source was really important for them. We're working hand in hand and different institutions are at different levels of maturity, so continuing to work with them to drive the adoption of UpToDate AI has been a big part of our focus.
Julie's team has been spending a lot of time there. Let me cover the second one and then Julie kind of fill in the gaps from there as well. Your question on sort of how we're working with AI scribes. And as you probably saw, we have several announcements, partnerships with AI scribes. We think it's important to integrate deeply in the workflow. And we're going to continue to drive greater value by working with those AI scribes and other emerging workflows, as well as with EMR players. So our goal is, as I mentioned in our presentation, is getting the right information to the right clinician at the right time and the right format. We're about driving greater ROI. And I would go back to that statistic I had in my presentation.
We're roughly 30% of the time a clinician uses UpToDate, they change their decision, they follow a treatment plan. That's real ROI to institutions. So the more we can embed that, reduce the friction of clinicians getting to that answer, there's real productivity value in that speed to answer. And then if you think about, you know, downstream changing decisions, that has real impact on hospital outcomes, length of stay. Those are examples. So driving clinical decision, lowering friction, driving it deeper into the workflow to drive downstream outcomes, that's where we're going. And those scribe announcements and partnerships we have are good examples of that strategy. But Julie, jump in if I missed anything.
Yeah, sure. I would just add one insight on the governance process with our customers. It is a fairly nascent kind of muscle in health systems, but also critically, critically important.
And so what has been a big effort on our part, but very important, has been collaboration. We've really taken a consultative approach to help organizations think about trust and safety and compliance in the context of launching an AI tool widely across all employees, clinical employees in the health system. And it has yielded fantastic results. And I considered, I think it will continue to be consultative in nature as the market matures on the health system side.
And then on the question about the pricing and monetization of UpToDate Enterprise Expert AI, you know, the question George was asking about how are we the decision to roll that into normal price increase. Maybe Stacey, you can address that and also comment on how the costs, the variable costs might be driven by usage of some of these tools. If you could start with that.
Maybe Alex, you can supplement on how we keep control of these costs.
Yeah, sure. As I mentioned earlier, when we have these enterprise kind of platform product lines, we typically tend to price based on value. So we have subscription-based pricing. And with UpToDate Expert AI, we are folding the Expert AI version into the overall offering. And as Greg described, this is a full, you know, platform where customers, as they move to the enterprise edition, they tend to also purchase the additional modules, including the drug information and patient education information. And so it really becomes a very, you know, negotiated price based on those factors, what upselling they might be doing along with the, you know, the number of sites they have and the number of practitioners.
Every year we go through and really look at the productivity that our customers are getting from our solutions, particularly when they expand their reach of the offerings. And we price accordingly based on that value. And then Greg, I don't know if you want us to talk about the specifics, but again, that's the approach for, you know, how we make sure we are monetizing the value that we're delivering to our customers. And that's the case with that UpToDate Expert AI as well.
Yeah, that's good. Yeah, Meg, I can add a little context on just the cost, if that'd be appropriate.
I think so, because I also have some emailed questions about how we control the variable costs.
Absolutely. Yeah. So costs, you know, are variable and depend on the complexity of the workflow.
I think Greg alluded to, you're talking about multiple LLM models and the UpToDate Expert AI, 30 plus agents and, you know, potentially hundreds of interactions, right? So this is quite complex. So first of all, we work with our hyperscaler partners to select the right models to help manage costs. But more importantly, we've made the investments in sort of our enterprise-grade offerings, our microservices, our API, our cloud-first strategy. They really allow us to, you know, focus on the information that we need at that point of care. Rather than overwhelming the models with a whole bunch of information, we can carefully select the right information that we need. And that dramatically lowers the token cost and it improves the speed. So those are a few ways that we're managing it.
Right. Yep.
As I mentioned earlier, you know, if we find that the, you know, the productivity and the use of the AI solutions increases, you know, we do have experience in doing hybrid models where, you know, there could be tiering of use levels for these enterprises to ensure that we capture the value and are, you know, have the appropriate levels of use accommodated in our pricing. Greg, you may want to also comment on UpToDate Pro because I think it's illustrative of how we price to, you know, we differentiate pricing depending on the segment of the market and what we're trying to accomplish, whether it's, you know, penetration for more scale or value capture. I think that's a good example.
Okay. The one example is around UpToDate Pro Plus, which was a unique package that we launched only a few weeks ago in the marketplace.
It's a combination of UpToDate Expert AI, so the technology advantage of an AI interface, plus some value-added content. We're finding a really nice uptake on that package. We're finding about a third of our renewals are choosing to upgrade with a 20% increase to that package. We're still early stages right now, but it's a really nice indication of if we're working with customers, put together a package that provides value, that we can see that in the marketplace and grow the business based on that value.
Okay. I have a couple of questions coming through the website. If George, if you've covered your questions.
Yeah, I mean, I don't want to dominate. I just want one final one. I mean, we talk a lot about the shift towards conversational AI as a search mechanism.
I'm guessing there will still be a decent tail of search queries or the way that people want to interact with the tool is the more traditional way of chaptering and looking through topics. I mean, to what extent do you see that as a pretty fat tail in terms of still existing in a few years' time or at least a competitive differentiator as well?
Yeah, no, what I would say, and then Greg, feel free and Stacey to chime in, but at the end, you know, sort of one of the core enduring strengths of Wolters Kluwer is the proprietary data.
So, you know, even as, you know, customers may want still that conversational search, and this is true not just in health, but in legal tax and the other verticals, they always want to be able to go back to, you know, the substantial, you know, content and data that can validate what decisions they want to make. And I think, as Greg mentioned before, this notion of traceability and being able to audit how they came to a conclusion is very, very important. These are high-stake decisions that they have to get right. And so, yes, I do think, you know, you know, customers want the natural language. They want that conversational thing. But at the end of it, they also need the combination with the substantial validation of what they're doing.
And I would say that more and more, it's sort of the customers are doing the research.
They, you know, want productivity and how they get to the right insights and answers. But with our AI and agentic capabilities and the fact that we're integrated into their workflows, the agentic tools allow us to move into productivity tools around improving outcomes. So it's really, you know, we believe this at, you know. This is a part of the workflow and the conversational interfaces, as Nancy says, with these, you know, with the backstop of the trusted, you know, ability to trace, particularly in the professions we serve, is so critical, but then also tying into the workflow to be able to deliver the results faster. You know, you certainly saw that in the collaborator example that Cathy showed in Julie's demo as well. So it's really the value of the entire solution, George.
That's great. Thanks very much. Great.
Can I post some questions I'm getting from the webcast? I'm wondering if both Jason and Greg could comment on how renewal rates have been trending this year. And secondly, can they just comment on why can't startups just hire experts and replicate what both of you do? So that's if both of you could comment on that. And then I have another question after that from the web. Starting with Jason.
So, as a market leader, we've got a deep relationship with our customers. And our renewal rates are very sticky. In fact, we see improvement as more customers choose cloud-based solutions. Retention rates have generally been in the mid-90s. We're seeing consistent uptick in cross-sell and upsell as customers move into the cloud.
And because it's a cloud-native platform and connected around what we showed as that wheel, it becomes very simple for customers to take the next module. So if they start with tax, they can move to document or practice management or any of the other Expert AI like Client Collaboration, which all work together around the core of that core platform. Yeah. You want to do renewals first? So we continue to see very strong renewals, as I said in my presentation, north of 90%. That's been consistent over the years. And as we look into next year, we see strong renewal in the future. It's really based on our value wheel that we described before.
Having that integrated solution of diagnosis, treatment, drug information, and education material all into one platform with our embedded nature into AI, our AI solutions, and into the ambient scribes, that full value wheel is really resonating. And that's helping to continue to lock in renewals and provide growth going forward. If you want to switch to the other question, Meg, I can pick that up and then hand back to Jason. As I mentioned before, our editorial process has been developed over 30 years. It's really rigorous. And while the slide showed the headlines, below that is a very, very well-matured process that allows us to have the level of clinical quality that we deliver in the marketplace. And it's leveraging those experts, those experts in each of the medical specialties, but also the processes that we've delivered and developed over the years to curate that content.
So we come out with the absolute gold standard for clinical decision support. And now with the two extra layers of clinical reasoning on top and validation levels, it's even a deeper mode that we built. The technology is one element, but bringing in our clinical experts to help develop those different stages of expert in the loop is really, really a differentiator.
Yeah. And I think one of the things that maybe, Stacey, you can elaborate that I think is so powerful is, you know, if you think about your own visits to your doctors, you know, they're spending that visit in the EHR, right? They, you know, they're literally looking at that screen. And we are deeply, deeply embedded in that. And everything that Greg described is that we're extending that integration, you know, across the hospital enterprise. And that's really, again, supports the renewals.
You can see that through the nine months, our recurring revenue is up 7%, which I think speaks to the strength of the renewals and the new sales of subscriptions across the enterprise.
Yeah. Absolutely. We see that, Nancy. You know, just to kind of elaborate that point, you know, when it's the clinician along with the care team, right? So it's a full integration across the workflow of those care teams in order to deliver deeply embedded within the workflow and across the care team. So really supports that important tie-in and delivers really solid productivity gains in addition to the best high-stakes, you know, the answers that serve them well in this high-stakes diagnosing world.
And just if I can pile, you know, on that point and then maybe ask Cathy or Jason to elaborate.
I think, you know, not only are the experts that we have again across the company so critical to ensuring that trust and accuracy and efficiency in search, which Alex talked about from a cost perspective, but the second thing is that we have this very, very deep knowledge of customer workflow down at sort of what I would call the micro task level, and as you think about agentic technology, which is really at the cornerstone of Stacey's strategy going forward, that leverages, you know, not only all the investments we've made in cloud, but that deep knowledge, and I think that that is where you are going to, again, see, you know, not only more growth across the customer base in terms of the tools that they adopt, but also this productivity benefit that I think the market is highly, highly focused on.
So Jason, I don't know if you or Cathy want to just give an example of how we're leveraging the agentic within tax and accounting.
I can start and then Cathy can add on. But, you know, it goes back to where Greg was talking about it. Similarly, in tax and accounting, we have decades of regulatory interpretation that are codified in a layered cross-jurisdictional logic that is used to build the proprietary content. And what's really critical is because all of this unified data is in a core platform like CCH Axcess, it is client-aware. So that is very unique in the way that we're able to deliver information that is connected to the specific compliance.
What Client Collaboration, what Cathy showed, is how that comes to life in an agentic workflow where you can take the interpretation of proprietary content, match it with the specific unique needs of each individual unique client, and serve that up. We can do that over and over at scale for each individual need for each individual client.
Yeah. Just to also build on that, the unique position that we have is that we have an open integration cloud platform that supports end-to-end, whether it's through capturing that data. Then, as Nancy was saying, we really have decomposed those customer workflows across whether it's audit or tax workflows, which have APIs available throughout the entire workflows to enable the agentic workflows.
Just recently, we launched at our user conference many different agentic solutions to enable you to capture data, understand that data, whether it's structured, unstructured, and to be able to then flow it into the tax return as an example, even agents for audit as well to really help you interact with the documents and help you understand what you need to consider as you're doing the audit. And that's only possible because we had that deep understanding of the customer workflows with the APIs across our entire ecosystem.
And I would add that because of the investments we've made into our cloud solutions, that whole process that you just described, Cathy, for your customers is a very, it's a superior user experience because it is seamless.
It really, you know, speaks to the integration that makes it easier for us to launch, but also seamless experience for our customers.
Also seamless for our customers' clients, which is also very key.
Thank you. I've got another question from the web here. This is for you, Greg and Julie, if you want to add on to it. It's really about, you know, how are you able to increase prices for UpToDate Pro Plus or UpToDate Enterprise Expert AI when you've got a competitor who's offering a product for free and advertising funded? Can you describe the dynamic there and what you hear from your customers about ad-funded, you know, options?
Let me start with price first and then jump in, Julie, as well. We can talk about the one example that I mentioned before, which is UpToDate Pro Plus.
I think it's a very good example where we're hearing from our clinicians, our users, that they really are valuing not only the AI experience, but the bundle that we put together and the positive feedback when we were developing the product to help drive that package itself. So we're hearing positive feedback on the trusted content that it's trained on is really important. And so this is a high-stakes environment where the risk of a hallucination is just something that they really can't accept. And so the fact that we put the same AI interface speed to answer, but it's trained on trusted content that they've used for years is a real differentiator. And then, as Julie mentioned, and as you saw in the demo, there's aspects that are really important to clinicians.
So, understanding the assumptions, understanding that, you know, this treatment plan is for a pediatric case, not an adult case, and all the other considerations when you're delivering care, it's very unique. Our deep domain expertise allows us to deliver that. Those are differentiators, and in a high-stakes market, there's high value to that. And we've seen that now in UpToDate Pro Plus and the uptake of 20% bump. But Julie, maybe you want to comment more on what we're hearing from customers and sort of our ability to deliver.
Yeah, I would just add maybe one nugget on the enterprise side, and that is our health systems are looking for an enterprise-grade partner.
That's what we hear time and time again, whether that is being able to support them on the governance process and the checks and balances, whether that's the customer success teams that come in and train and teach clinicians how to use AI, whether it is the ability to work with administrators on ensuring that any tool they use does not contribute to deskilling of clinicians, whether it is the security and validation we put around it. Ultimately, those things together make us an enterprise-grade vendor, which is truly what they're looking for.
Ma ybe, Julie, you could just talk. Sorry, Stacey, maybe you could talk also. Yeah, about advertising. Yeah. Yeah. While Julie was, you know, talking about the enterprise, I would just add in we have long competed against lower-cost solutions going into the enterprise.
To all the points that Julie and Greg raised, even during COVID when, you know, budgets were very tight, we have been successful maintaining strong renewals, strong upsells. This is kind of standard practice for us. All of the, you know, for all the reasons that Julie and Greg described, we continue to be very successful. Yeah. Maybe, Julie, you could, yeah, just talk about the advertising risk there that because Meg specifically asked about that. Yeah.
As you know, 80% of our business is at the enterprise level. And so an advertising model at the enterprise level is just, it doesn't really work. And the other thing I would say is our customers, a lot of them use UpToDate and receive CME credits. And so an advertising model, you know, is more, it's more complicated with that model.
They want to be sure that they can get ACCME accreditation through a subscription-based business.
Great. Thank you. I've got two final questions from the web. Then I think we can call it to an end. I'll pose them both. Maybe Stacey and Greg can comment on these two questions. One is just general, what is happening with usage in UpToDate and perhaps also generally with information solutions? You know, what are the trends and where do you see that going in the future? Secondly, could you talk a bit more about your partnership with what we're doing with Microsoft, Microsoft Dragon Copilot? You know, what should we think about with that particular partnership?
If you want to, Stacey, I could take the usage. Then, Julie, I might start the Microsoft and then hand it over to you to go in depth there.
And so for usage, we know that our users are experimenting with new LLMs and MLMs, and that's expected. One thing for UpToDate is early on, even before LLMs really kicked in, we designed UpToDate to provide greater efficiency for our clinicians. And we expected usage to be impacted by that. So we designed clinical summaries and key points panels that will get our clinicians to the answers faster. So, you know, the usage impact is not a surprise to us because even before LLMs kicked in, we saw our users becoming more efficient. And we see that as a move from volume to value. They're getting more value out of it. It's a productivity tool to get them to the right answer faster. And as UpToDate Expert AI penetrates more and more, that trend is going to continue.
So what we're looking at is how we can deliver real ROI downstream and real productivity gains. And as we're more integrated, we reduce that friction from when the clinician wants to get to an answer to that answer itself, deeply embedded in workflows through UpToDate Expert AI. We're going to continue to add more value for our customers, and especially at an enterprise level where we're more deeply integrated at the enterprise level. And that's going to continue to drive our growth going forward. On the Microsoft side, we've had a long-term relationship with Microsoft. It's been a very productive one. We're seeing some really good experiments with them over the years. And our collaboration with Microsoft is really going to help drive that strategy of being more deeply embedded in the workflow and providing greater outcomes to customers.
Julie's been right in the mix on this, so I might give her an opportunity to kind of chime in on where we are with that relationship. But it's a trusted relationship that we're really looking forward to the impact.
Sure. I can give some more color on the use case. And the use cases that we're working on, we're working very closely with the Dragon Copilot team as well as with joint customers. And we have a large base of joint customers to source kind of their priority needs. As customers are investing and have been investing significantly in ambient, they're asking us to ensure that their gold standard of clinical decision support can be truly incorporated into that kind of new priority workflow. So that's been our focus.
One of the things that's really, really exciting about the partnership is the ability to provide what I would call context-aware guidance. And so what you have is you have this new highly valued workflow, the ambient workflow. What it also includes is information specifically about that patient at the patient encounter. You marry that up with a gold standard of decision support, and you can drive meaningful impact in care. So that's really where we're focused. And just from a customer perspective, we have a number of customers lined up in terms of being our sort of alpha group there. Lots of excitement about the opportunities this unlocks.
Great. Thank you. Operator, are there any more questions on the call?
There are no further questions on the phone line. So I would like to hand it back to your host for some final comments. Okay.
Thank you, everyone, for joining us today. All very good questions. I hope we answered some of your AI questions, and if you've got further ones, feel free to reach out to us, and we will try and help you out. Thank you very much for joining us today.