Welcome. I'm Cynthia Hiponia, Vice President of Investor Relations, and I want to say welcome to our webcast attendees, as well as our analysts and investors here in person. We've got a really great afternoon scheduled for you guys. Aaron's going to kick it off with an overview of our corporate strategy. Ben, our CTO, is going to do a deep dive into our AI strategy. Diego, our Chief Product Officer, will go over our product roadmap. We'll take a short break, and then we have Olivia discussing our go-to-market strategy. Mark will do a deep dive into our sales. We are very thrilled—you may have seen our release this morning—to announce a strategic partnership with DataBank, and we're excited to have Matt Charlson here in person. We'll do a nice interview with Mark and Matt after that.
Finally, Dylan will wrap it up with an overview of our financial strategy, and then we'll have Q&A. A few housekeeping things. Forward-looking statement: just know that the presentation contains forward-looking statements, and these statements are regarding our future business and financial expectations. There are a number of factors that could impact this, and we understand you have no obligation to update them. We encourage you to look at the Risk Factors in our SEC filings for more details. During today's presentation, we'll be sharing information on our roadmap and future products that we are planning but are not yet available. The information is intended as a guideline on our product development map and should not be used in making investment decisions. Lastly, unless otherwise indicated, all metrics are in non-GAAP. With that, I will kick it off to Aaron Levie.
Cool. Thank you. Thanks, Cynthia. Welcome, everybody. We're going to walk through a little bit of our overall strategy at Box and obviously how AI is transforming our market opportunity, and then very excited to dive into a little bit of how this translates into what we're building as a product and as a platform. Our long-term strategy for driving profitable growth at scale at Box really breaks out into four big components, and this is sort of what we think about internally as we think about driving our strategy. The first is we're going after a $100 billion+ market in the next few years. When you think about the total TAM that goes into content management, the infrastructure around content management, the next set of applications that we can serve given our efforts in no-code apps and AI. This is a massive market.
It's continuously growing, and we're seeing new growth vectors driven by AI that we'll be talking about shortly. To get that market, we have to build the leading intelligent content management platform. It's driven by AI and lets us really bring intelligent workflows to the enterprise. What you'll hear from Ben and Diego is how we're building out this platform opportunity, what our multi-product strategy looks like, and then Mark and Olivia will bring this home in terms of how we deliver this to market with our Enterprise Advanced offering, as well as new consumption-oriented models driven by AI and our API business. To bring Enterprise Advanced to market, this is really about continuing to really get directly within our customers' environments in key verticals, in key segments, and industries. You're going to see us bring along, as well as work with the partner ecosystem to drive this.
We're excited to have this partnership with DataBank as one great example of what that looks like. You'll hear a bit more about that shortly. You bring this all together, and if you look at how we're running the business with a high degree of rigor and operational excellence, this allows us to really drive double-digit top-line growth and deliver on our rule of 45-50 target in this sort of medium-long run from a financial model standpoint. These are the long-term sort of core strategic areas that we're focused on in terms of driving our strategy, and we're going to dive into all of them today. At Box, our mission is to power how the world works together. When we started the company, it was really a simple but powerful idea.
It was that you should be able to work from anywhere, with anyone, get access to all of your data and information. What we quickly saw was that this was rippling through the enterprise. It was really changing how enterprises were collaborating, how they were working with their information, and how they were really leveraging their unstructured data and content. Today, we're incredibly excited to be able to work with a significant portion of the Fortune 500. We have 120,000 customers on the platform today. We work with many of the leading brands across, obviously, essentially every single industry that is out there. The thing that I get most excited about is just the breadth of industries that we get to serve. We get to serve everything from financial services institutions to technology companies to media and entertainment organizations to professional services firms.
This ranges from the small startups that are growing rapidly to some of the world's largest companies that have scaled on our platform. Just to take a few examples of companies that we get to work with every single day, we work with Block to be able to help them secure some of their most sensitive data. If you think about any kind of financial technology company, the amount of compliance that they have to be able to orient around, the amount of data that they have to work with is obviously incredibly important. We get to power how they work with their information. Companies like Boston Dynamics that are at the forefront of robotics and advanced technology leverage Box to be able to collaborate securely on their most important information. We also get to work with leading industrial companies and automotive organizations.
GM, on their defense division, obviously works with incredibly sensitive data that is in a highly regulated part of their business, and we power how they collaborate and share that information. Globally and internationally, in areas like Japan that we've talked a lot about the success that we're seeing, we get to work with amazing institutions like Toyota that are at the forefront of advanced technologies, electric vehicles, self-driving cars, and more. What every single one of these companies have in common is their businesses run on unstructured data. This is how they operate. This is how they share. This is how they collaborate. This is how they automate their organizations. We're incredibly excited about what we've accomplished. We're very proud of the kinds of customers that we get to work with every single day.
We're at a moment right now where the opportunity is substantially larger than we've ever seen, and it's really causing us to almost think about Box in a very different way. We're at a moment where everything is changing about software, everything is changing about technology, and we're really just at the beginning of what this journey looks like. When we think about the future of business, there's very clearly an emerging concept where businesses are going to be run very differently in the future. We're entering really an AI-first era of business, and we think this is going to transform how businesses operate in the future. The first is that we have AI agents entering the workforce, and this is going to have some really important implications. It's going to change how we get our work done. It's going to change how we collaborate.
We have the ability now for any employee in an organization to deploy AI agents across a wide variety of tasks that they perform. This allows us to either augment the work that we're already doing or begin to automate and solve problems that we weren't doing before. A lot of the upside opportunity in AI is not just replacing what we're currently doing with automation, but actually going after problems that we just were not spending technology time on or human energy on before. We believe AI agents are going to drive a massive productivity shift in the enterprise. You're going to hear from the team today around how we're using AI within our organization across nearly every function and what that has done to our productivity. We're going to see that this really shapes how work is happening in the future in every organization.
As you have AI agents enter the enterprise and they really sit alongside us in the workforce, it means we can begin to automate nearly any workflow. We can take really any kind of business process. It could be a contract management process. It could be a digital asset management process. It could be a clinical trial workflow. Any workflow that today is very manual, very labor-intensive, it deals with a lot of unstructured data, so it's sort of hard to automate previously, we can begin to automate these workflows. We're seeing that this is going to be a massive shift in how we think about workflows in the enterprise when you can deploy AI agents against them and begin to actually streamline these efforts. We also begin to have the ability to get instant intelligence from our data.
This is one of the things that has been really eye-opening for me is I'll have conversations with CEOs, CIOs at a level where these leaders were not thinking about their unstructured data as a source for differentiated sort of business value previously. If you think about maybe comparing today versus three or five years ago, five years ago, you think about content, you think about records management. I'm going to go put my records somewhere. Maybe we'll get back to them if there's a legal request or there's some kind of business process where I have to pull that data out of a system. You're sitting on this incredible wealth of information that you're just accumulating over time. Now, all of a sudden, AI is letting us actually tap into those insights and tap into that data. Enterprises are saying, "Wait a second.
What information do I actually have inside of this gold mine of unstructured data? It is causing them to really think about, to rethink the value of that unstructured data in their enterprise. We will get into examples of what that would look like. As you imagine, we are automating workflows. We are getting new insights from our data. We can have agents enter the workforce. It also means that customers are going to have completely new expectations from the companies that they work with, which then creates yet another major trend in an AI-first business, which is you have to modernize your product experiences so your customers actually have a better experience. There is no way that in three or five years from now, you are going to expect the same kind of slowness when you are onboarding to a new company or when you are getting a customer support question answered.
That means that companies have to be wired up to deliver these AI-first customer experiences, which again means your workflows and your data has to be in a good position. We are very clearly entering this completely new era of business. It's going to change how we run our companies. It's going to change the value propositions that companies have for their customers. It's going to accelerate breakthroughs in essentially every single industry. What this is also going to mean is clearly that we are going to need a new era of software. Software is fundamentally going to change to be able to deliver on this potential of having an AI-first business.
If you think about maybe the biggest buckets of software that we've had, the eras of software that we've had, there's effectively been two before now, and we're really entering this third era of software. The first was we had systems of record, and this was really born out of the client-server era of technology, the PC connecting to the server. These were mostly back-office systems. They're on-prem. It solved our initial ERP use cases, some of the initial CRM use cases, ITSM, and so on. These were largely back-office technologies, and they primarily worked on your structured data. They could really only automate things about your business that you could put into a database. This was the era that Oracle, SAP, Siebel, et cetera, kind of were born out of and grew up in.
These technologies are still today incredibly important, but we kind of know how they work, and we know what we can automate. It is really a back-office set of functions that we can deliver. We have this era of systems of engagement, and this is kind of the era that Box grew up in initially. It was the era of the Salesforces, the Slacks, the Zooms, the Boxes, which was really end users inside of an organization being able to work together. This is how you would communicate. It is how you shared. It was really cloud-based, often mobile, and it was much more unstructured. We were sharing documents. We are sharing conversations. We are doing a conversation. We are communicating. Much more messy, much more sort of free-flowing, but it hit every employee in the organization.
All of a sudden, the amount of people that would touch software kind of went up by an order of magnitude. This drove the growth of the cloud. It drove the growth of mobile, and it really has reshaped how we work. Today, everything about how we communicate, how we work is 10 times faster than it was in the era of systems of record where you do something, you put it into a database, and maybe you query it later. Now we're entering really kind of a new chapter of software, and it's really systems of intelligence. This is an era where not only is it going to build on the prior two eras, but there's a real shot that it actually sort of subsumes these categories of software as well.
There are some really interesting debates happening in software right now of what that looks like. It is an era where it certainly builds on the existing infrastructure that we got with the cloud, but it is really defined by AI. It is an era that is going to impact the front and back office as well as the customer ecosystem. It reaches and spans every single section of our business. It allows us to take the unstructured, messy work that we were doing in the systems of engagement and the structured work that we were doing in the systems of record and unify that. All of a sudden, our unstructured data becomes structured. We can query it. We can ask it questions. We can summarize information. We can get a sense of everything going on in the organization that was simply not possible before.
Ben will show some examples of what that looks like with all of the unstructured data that we have. It is a technology category that really is going to be defined by AI agents. We are going to have AI agents span really almost every single use case in the enterprise. It is even causing questions in the software industry of what is the future user interface in software? How much is going to be in a GUI? How much is the agent just doing behind the scenes? No matter what, we are very clearly going to have this new era of systems of intelligence. It will talk to our systems of record. It will talk to our systems of engagement, and yet it will also be its own sort of category potential as well.
Why are we really, really excited about this, and how are we defining the company in this era of systems of intelligence? It's because these systems of intelligence, they fundamentally thrive on all of our unstructured data. If you look at the growth of unstructured data the past couple of decades, it's astronomical. We are creating just an incredible amount of unstructured data in almost everything we do. Every video that is recorded is unstructured data. Every PDF that is generated from a bank system is unstructured data. Every contract that we write, every conversation that we store, every piece of log data that we save, it's unstructured data. This is really where the growth is coming from, is in all of this information. What's amazing is that for the 10% of our data that is structured, we've always been able to query it.
We've always been able to summarize it. We've always been able to automate the workflows around it. Yet the 90% of our unstructured data that we have, we've gotten a substantially less amount of value from this data. This is, again, the challenges pre-AI that we had. We can't really tap into the insights of what's sitting in all of this unstructured data. What we get really excited about is that it's 90% of our data, and we actually believe it's probably our most important data. If you think about that intro video that we showed, whether it's delivering a new cancer treatment, whether it's taking a rocket to space, whether it's closing a multi-billion dollar acquisition deal, all of this work runs on unstructured data. It doesn't run on structured data. It runs on unstructured data.
This is the lifeblood of so many of our most important business processes. It's how we launch products. All of the R&D data, all of the roadmap information, all of the documentation, that's unstructured data and content. The ability to onboard new employees, recruit them, get them into the right job, that's all content and unstructured data. The ability to look through an incredible trove of clinical lab results, be able to get research information and insights, that's unstructured data. The ability to onboard a new applicant in a loan process, make sure that they have all the requisite information and do a KYC workflow, that's unstructured data. The ability to quickly work through an insurance claim, all of the media that goes into that, the images, the videos, the form information, that's unstructured data. All of this today is untapped and underutilized in an enterprise.
We talk about this a lot, about how did we get here. First of all, we got here because we did not have AI. We could not have actually really, truly unleashed the potential of this data before. We also got here in terms of not being able to deliver on this value proposition because right now in most organizations, their data is fragmented across a wide variety of systems. We have a bunch of silos of where this information goes. It goes into workflow technologies. It goes into legacy document management systems. It goes into end user collaboration tools. Each of these technologies, sort of, they were born out of a reason that the organization adopted it. They needed to share a file or have a secure data room or do an onboarding workflow.
The problem is that most organizations now have this highly fragmented environment. The systems are disconnected. The data is really not as secure as it needs to be. It's way too costly because we're restoring the same data in multiple places, which is obviously wildly inefficient. Worst of all, when AI now comes into the picture, you're not going to be able to get the full value of this unstructured data in your enterprise content in this kind of architecture. It's going to be very, very difficult to pull off. We know that there's a better way, and this is what we've been working on at Box.
We have been building this platform to really solve the full continuum, the full lifecycle that content goes through in an organization, the moment that it is created and generated, to how you collaborate, to how you share, to how you access it from anywhere, publish it, automate a workflow around it, get an e-signature, or archive and govern it at the end of the lifecycle. We have been building that platform. The key to that platform is we put content right at the center of everything we do. Content is the core, sort of the core kernel of what we operate on, as we have a platform that manages the content. We then have all of the capabilities for the full lifecycle of that content. Most importantly, we put AI right at the center of that platform.
We didn't think about AI as sort of this afterthought capability that we think about as sort of the separate appendage. AI is central to the entire platform offering. Then we integrate across the software landscape that our customers work with. We want to work with and embed into all of the technology that they have. This is why we have over 1,500 applications that we integrate with. We integrate into IBM's technology. We integrate into Salesforce. We integrate into Slack or Teams or ServiceNow. No matter where an employee is working from or where you're executing a workflow from, that system can talk to the documents and content in Box, and you get a single file system. You're not replicating data. You're not fragmenting it. You're not restoring it in multiple places.
Then most importantly, all of this functionality works within that application as well. Any workflow, any process you're doing in one of those systems automatically will work with the data that's in Box. That was what we've been building out. We added a new component to this mix just in the past sort of two months in GA and sort of four months since we announced it, which is Box Apps. With Box Apps, now on the other side of this, we can support any of our customers' custom use cases around content. All of a sudden, all of those applications you used to have to buy for every single team or department in an organization, the Box platform can now deliver with Box Apps. I need a digital asset management system.
You do not need to now go replicate your data and have another technology. I need a contract management system. Similarly, we can deliver that type of use case. I need an invoice processing technology. We deliver that. You can see that there is this almost gravitational pull with your data and unstructured data, which is that the more that you have the functionality around the content, the more that you get the intelligence around the content and AI, it makes it easier to then integrate different applications and systems on top of that content. That is the power of having an intelligent content management platform. You store the data once. AI operates on it.
You get all of the functionality in one integrated platform, and then we connect to all of your applications, whether they're Box-built in our no-code application environment or third-party applications that we integrate with. This is what this platform looks like a little bit more blown out. Diego and Ben will share a bit more on this, but it's really about the kind of building up that core platform in each of the core layers that we've invested in. Our global infrastructure that's nearly infinitely scalable for any use case, a data protection layer to help our customers protect their most important assets, content services, which lets us automate workflows and build no-code applications, and then our AI platform, which is what we're investing so much in.
With that platform and with our intelligent content management set of capabilities, enterprises can fundamentally really re-engineer and unlock the true value of their content. There are three major areas that we're investing in around AI. Again, Ben will share a little bit about what this looks like. The first is that we can help companies just generate instant insights from their data. We've talked about this a little bit over the past couple of years, but this is sort of this idea of having retrieval augmented generation on your own data in your environment with no code. Products like Box Hubs from Box let you instantly have a portal to ask any kind of question from your unstructured data. You can instantly get insights from any of your unstructured data. The second is that you can pull out intelligence from your content.
If you think about what's hidden inside of all of our documents and PDFs and media assets, it's all of this structured data that is right now in an unstructured form. The name of the client that you're selling to, the due date of the renewal, the amount of the deal, key clauses that you really have to pay attention to. Right now, that looks like any other piece of data inside of a document. You can't really automate a workflow around it because you don't know really all of the normalized versions of this data across a large number of documents. Now with Box AI, we can pull out the structured data from that unstructured piece of content, put it into a database that we have in our platform, and then you can begin to pull out intelligence and insights from all of this information.
This really ties into our no-code apps that we launched as well. Finally, if you can pull out the structured data from your content, it means you can automate your workflows that previously were not possible to automate. If I do not know what is in the loan agreement, if I do not know what is in the contract, if I do not know what is in the bank statement, I cannot automate the workflow around it. All of a sudden, if I can pull out the structured data from that content, now I can actually streamline and automate those workflows. AI lets us automate way more workflows than were ever possible before. This is where we are going to talk a lot about intelligent workflow automation in our platform.
When you add it all up, AI is really going to transform what we can now do with our enterprise content. We see this as a massive expansion in the kind of use cases that we can solve for our customers. We have traditionally talked about the opportunity of Box as going after the knowledge workers inside of an organization. You will see from Dylan, as well as just in general, what we talk about is we have a certain amount of seats that we have sold. We have seats that we could sell into those same clients, and then we have seats that we can sell into similar companies that are technically new logos in the future for us. There is now another sphere of that opportunity, which is the ability to apply AI to work.
This is really the AI agent opportunity where you can see a massive expansion in the kind of TAM that we can go after. I'll share just one kind of anecdotal example of what this looks like. We're increasingly having conversations with customers where maybe they are tapped out on seats that we could sell into that organization, but all of a sudden, they want to go and automate a workflow that is today not represented by the seats in that company. Maybe it's an external firm that they were going with, or they were just not even working on automating that process previously. All of a sudden, AI agents opens up another category of TAM that we can go and sell into.
Customers coming to us saying, "Hey, I'd really like to automate that contract workflow," or, "I have a lot of insights in this massive mess of unstructured data. Can I deploy AI against that, understand what's in that information?" In those cases, we're not cannibalizing any seats that we've already sold to the organization. It's net new spend for that company to go and automate a process. We see this as really expanding the total addressable market, not just for Box, but really for software in general. For us, we're going to be going after this in really a consistent fashion that we have over the past decade plus as we've introduced multiple plans with one new addition, which is AI units that lets us go after more of these consumption use cases.
We're going to continue to introduce new plans, and this allows us to go and drive average contract value up, get deeper into the customer workflows that we're focused on, which means that we're much stickier and then gives us more revenue potential over the long run. When we founded the company, it was a really simple, single core offering, make it easy to share and collaborate on data. Then we expanded into add-on and sort of bundled seats. That was our Enterprise Plus offering. Now with the launch of Enterprise Advanced, which is just a couple of months old, this lets us now expand into all of the intelligent workflows in the enterprise. Going from just secure content management now to intelligent workflow automation with the additional capability of having a consumption model with AI units.
It is not just seats, but lets us actually deploy AI units for those higher-end consumption use cases. This is how we are going to continue to monetize the platform that we are building out. Olivia will double-click quite a bit more into some of the details on this, and you will hear from Mark similarly on that. This really drives that double-digit growth that we are focused on achieving. Long-term strategy, go after a massive market. The market is only getting bigger by the day for what we are going after. Build the leading platform to actually go and tap into this market opportunity. Make sure that we can get to all of the customers and accelerate their transformation with their unstructured data, leveraging our intelligent content management platform. Do that with a level of efficiency and operational rigor with that focus on the double-digit growth rate.
That is the strategy that we've put together. To share a bit more about what we're up to in AI and to give a little bit of a sneak peek in a couple of areas that you'll see throughout this year and a little bit beyond, I'd like to invite Ben, our CTO, to share a bit more. Thank you, Ben.
Hello. I'm Ben, Technology Officer at Box, and I'm here today to talk about our AI strategy. Today we'll be focused on a little bit of a review of our current strategy in addition to speaking about what's coming up next, and in particular around AI agents and agentic workflows. As you know, as in further view, we've released a number of features around AI already.
We have things that are like our Hubs AI, which is for generating instant insights into your documents, into corpuses of data. We have here the ability to do retrieval augmented generation and the ability to go through and understand at scale what it takes to understand the key details of your documents. We have the idea of data extraction, turning unstructured data into structured data and then using that and storing that in our metadata system and being able to query it and filter it and to run at scale these kinds of structured operations. We have the ability to do this via API and via the ability to automate that inside of Box.
As we talk to customers, many of them are trying to solve these kinds of problems, and they're looking at ways by which they can build upon AI on top of all the capabilities that we offer inside of Box for your unstructured data. A couple of examples of how we're seeing this play out today in the market. We've had recently one of our largest technology enterprises. They've deployed Box AI wall to wall in their organization. This is one of the first AI deployments they've done across the company. We've also been working with a financial capital firm who have recently done a million data extraction documents. This was part of an audit that they had, and they needed to go through quickly and figure out a bunch of data and pull out some of the key details.
This not only saved them countless hours, but it also helped them reduce the risk in their audit. We have another example of a company that we were working with in the travel industry who had this very long manual process that they utilized for many years about when they're talking to their clients and they're getting these kinds of travel itineraries. Using Box AI, they were able to automate this, not only saving their travel agents a ton of time, but also being able to accelerate how quickly they can interact with their clients. These are just a few examples of the many different examples that we have seen with Box AI overall. The key is that we did not just build these capabilities for these clients. Instead, we took a platform approach for how we deliver this.
We have our layers of infrastructure, data protection, content services in our AI platform. You can think of it that the AI platform's job is to build on top of everything that we've done so far in Box and over the history of Box's content platform and then use this to then power our applications so people can use it directly inside of Box, also our integrations, and then also build on top of this as a platform whenever you're doing AI on your unstructured data. If we dive a little bit deeper into the AI platform, this is what we've been working on for the past couple of years where we've been taking not just any content, but almost all forms of unstructured content and being able to apply AI to them.
This is across things like different modalities, different file types, being able to process not just the text-based documents, but also the multimodal from images and then soon for audio and video. Critically for all of this is that no enterprise will ever adopt any of these AI solutions unless they're very confident about the security, the compliance, the governance, making sure that the AI access is controlled so it doesn't operate on things that the users have access to.
For the models, one of our goals is to support all of the top AI models available in the industry from companies such as OpenAI or Anthropic or Google or our Meta models, and to offer them on trusted platforms so that customers will be able to adopt these without having to worry about where their data is going and having to worry about the security of their data. We have tools and integrations that we are developing and that we use to power our integrations with our partners in addition to if somebody wants to build on top of our Box platform in an agentic way, we offer these new frameworks, capabilities. Across the board for our AI experiences, we offer a consistent experience on top of your data.
While we're building all this, we're very happy that the AI models themselves are evolving rapidly, and they continue to evolve seemingly almost every week. In terms of the way that for the use cases that we care about for the AI models, we've seen a dramatic increase in the way that they have been able to handle the kinds of questions and the kinds of challenges that we put in front of them. Meanwhile, the cost for inference is dramatically dropping. This means that for any given price point, we're able to do more AI and actually do AI faster because typically the speed increases too. The new models have new capabilities, not only being able to process images and other types of audio and video, but also the new reasoning or thinking agents are very helpful at dealing with very complex challenges.
As these models continue to evolve, we see, as Aaron mentioned, that when we talk to customers, there is definitely a trend towards the use of AI agents. When we say AI agents in this context, we are referring to not just AI models or the idea of an AI assisting you, but for AI doing more complex workflows and to be able to have an objective and to be able to have a mini workflow to go through and complete the problem or solve harder problems, oftentimes working as a team with other agents or with people.
This we think is incredibly important for the future because we see that over time, and we expect that not only will knowledge workers be able to call upon teams of agents to be able to do work for them, but they'll be able to then have AI agents handling certain problems that today lack automation and today take our very expensive processes. AI agents can add efficiency in addition to helping to change fundamentally the way that the enterprise knowledge workers work because they can then call upon these agents.
To prepare for this agentic future, we are continuing to evolve our AI platform in the background to take advantage of these newest capabilities, not only utilizing the latest models at all times to power these AI agents, but also building in the customizability for these agentic routines and these agentic workflows, and then importantly, being able to integrate this with the best-of-breed platforms that contain your enterprise data today. Digging in a detail for a moment, whereas today and for the past few years, we've been seeing a number of technology approaches that help provide the kind of capabilities that you've seen for AI around your content, things like retrieval augmented generation, data extraction, and so on. We are now evolving those into a world where you're doing agentic capabilities in the same way, so agentic RAG and agentic data extraction.
Some of the newest capabilities like deep research or deep search give a whole fundamentally new approach to being able to extract data. You've seen this on things like the way that some of the tools like OpenAI or Gemini help you do deep research on the internet. You can then do so for us, we're going to be providing the capability to do deep research on your enterprise content. With the idea of providing more customizability of these agentic routines so that you'll be able to customize the way the agents work within your org. I'll show an example here. This is an example that we were working with one of our customers, IDEX. Here in this example, they have a hub of some of their big events from the year. This is from the fiscal year.
They ask a question using our new proposed deep research capabilities here. It says, "Generate a report evaluating our fiscal year 2024 performance." Rather than just summarizing some specific bullet points, you can see here that the AI will think for a while. This is kind of the newer model, the capabilities around thinking more. Not only generate a text-based output looking through the text-based capabilities, but also look through the pictures inside of the PowerPoints, watching the videos, listening to the audio forecast, and then being able to generate insight into all aspects of this kind of question to generate this report from inside your enterprise. Historically, this has been very hard for AI to do because it's quite a complex, lots of data that goes behind it.
Now with these newer techniques around the way the models work in addition to the deep research capabilities, you see that this is now possible. Going through and seeing that it can not only reference the different resources, but also cite the videos, citing the audio, citing the images. This is a good example of multimodal AI. Something like deep research is useful for most users inside of a system like Box, but we think that a big part of the value will be when we start to have AI agents that are dedicated to specific tasks. Many companies work with a wide variety of these content-driven workflows internally. Varies by company and by industry, but things like imagine that you're dealing with reviewing contracts, and you might need help from an AI agent to specifically review a risk for the contract.
You have a specific dedicated agent to that. If you're in life sciences and you have a regulatory process you need to go through and you need help being able to analyze and to summarize the different data to be able to make it available to turn in for regulatory reports. On and on, almost every organization has many of these kinds of workflows that they need to go through related to their content. This is where Box will be providing these kinds of customized agentic workflows so that they can be able to use agents in their organization. We do not see that this is something that Box will do by itself. We see that we are part of an agentic ecosystem where many different platforms, including many of our partners, have a best-of-breed approach to being able to use agents in the enterprise.
A big part of the reason we think this is because when you're looking at enterprises, of course, one of the things that anybody needs, whether it's a person who's coming to the organization or who works there or these enterprise agents, either the ones that they build themselves or the ones that they buy from another company, is that it's critical for them to be able to do real work is to access the data in the enterprise. This data today is stored across multiple vendors, custom solutions, and companies like Box. For us, one of the keys is that to not only have these agents have the ability to go and reach into this different system to extract the key information for their workflow, but also to use the agents and the agentic capabilities that the platforms have available.
If you're building one of these workflows inside of a company, you'll have the ability to have the agent talk to our agents, our Box AI agents, which will help them navigate and figure out how to get the key information available for whatever the agentic workflow is that they have available. This is how we see the world of the agentic ecosystem evolving. I'll give you a specific example of one that we're working with our customers on around sales RFPs. We've heard this from many customers, and it's a common problem, which is that it's a really critical process internally when you're given these RFPs that come from a customer. Typically very long, very detailed, requires a subject matter expert to sit there to answer all these different questions.
Using the power of an agentic technique, in this case, this is an example of one that Box would offer, we would have the ability to go through, let's look up in already existing RFPs the answers to these questions, sort of like a library of these answers. If you can't find the answer because the question's new or about a new product, then go look up in the product documentation to figure out if you can answer these questions. Not only have the answer, but then double-check the answer, make sure that it's formatted properly, make sure that there's no discrepancies, make sure that it has the right tone and conciseness for this type of offering. Do QA on the RFP answers. Turn around and actually generate the filled-out RFP.
This is something that probably many, many organizations do constantly that takes a very long time. This is something that we can do in minutes with an agentic workflow. This is the kind of agentic workflow that we would offer. In many ways, one key piece of this is that this is not the only part of an agentic workflow these companies might be interested in. Probably you also have an agent who's dealing with the sales opportunity, working with your CRM system, working with the way that you handle sales internally, following up with the customer, creating the follow-up emails, and making sure that you have an end-to-end interaction with this type of process.
We see ourselves as a critical step in here when you're ever dealing with content workflows, but part of a bigger ecosystem of the way that we all work together as platforms and with customized AI workflows. In terms of commercializing, we have two different models that we use today for Box AI. We typically have been trying to provide for our customers and for our users the ability for them to use AI across their data. This is we consider to be just an important part of giving users the ability to understand their data like a modern content management solution. Also, we've also introduced this idea of having Box AI units. You can think of an AI unit as like a fundamental work unit for AI.
We use this for some of the more high-scale operations, like for instance, data extraction or API usage. We are able to charge basically for the amount of work needed for AI. We believe that between these two types of models, we are able to not only monetize the AI, but also be able to use these for future offerings going forward. With that, I am going to turn it over to Diego, who will walk us through our product roadmap.
Thanks, Ben. Thank you. All right. Thank you, Ben. I'm Diego Dugatkin. I'm the Chief Product Officer at Box. I'm super excited because the use of AI is really going to help us expand what we can do across the whole portfolio. In order to grow intelligent content management, we're going to basically keep applying AI across everything we do, as Aaron set out. Our priority in these rapidly changing times is to be the trusted partner to our customers to really find a content strategy that brings AI to everything they do, working with us. In these AI transformations, every department, every industry, every part of everything we do with content is going to benefit through the use of AI. AI is going to accelerate the ability to basically increase productivity and competitiveness across every industry.
In terms of which is the right partner, Box is the best partner for these transformations because we already have the way to protect, manage, and integrate more than one exabyte of truly valuable enterprise content. As Ben covered, the implementation that we have already built as an agile, secure, and transparent AI system is best positioned to accelerate that in the industry. Now I'm going to share how we're going to leverage that across the best-in-class AI capabilities for all our portfolio. Ben already covered the first part, which is how AI is implemented and basically the capabilities that we have already built for that. Next, I'm going to dive into our AI strategy for workflow and collaboration.
Now, in terms of how AI is applied, we have workflows and collaboration that will see rapid productivity improvements that Aaron talked about, and also the ability to process unstructured data at scale to deliver multi-agent flows that are going to truly transform many processes for many use cases, from claims to applications to drafting reports to deep research across the whole industry. Now, Aaron mentioned briefly Box Apps. We're going to talk about that in a minute, but it's the integration of Box Apps with the use of AI across the implementation of workflows that are going to truly transform what we can do with content. Starting with metadata that is at the core of many workflows, it makes unstructured data much more searchable and actionable at scale. Previously, extracting metadata was a manual process. It required basically bolt-on solutions and only worked in specific document types.
Now, Box is building the best way to extract metadata using the best-in-class AI and LLMs to extract metadata at enormous speed and consistency. We can work with a wide set of documents, document types, and images, and customers can already do this using our metadata extraction APIs. Soon, they will also be able to do that directly with Box Preview and Box Apps in the platform. This year, Box is continuing to invest in metadata extraction and to build on these recent releases. Customers can fine-tune AI extraction basically using extraction fields and prompts, and we're providing a full taxonomy support to match our customers' business language directly. We're also enabling automatic extraction at scale for large batch processes with hundreds of thousands of files at once.
In line with our AI principles, we're maintaining the ability for customers to decide when to keep human feedback and human in the loop for the whole process. Always keeping the ability to select when humans are also required to be part of it. Now, once you have extracted structure from unstructured data, you can also put that into action. We're upgrading Box Relay, our workflow automation tool, to handle large and more complex processes and power agentic workflows. Relay will now support also process automation at scale, complex workflows owned by different departments, teams, or enterprises, as well as custom workflows that are built using our APIs. Relay will also be the single workflow orchestration layer, whether it's advanced logic or workflows or support for non-linear or parallel execution for workflows.
Relay will drive it, for example, with content libraries or content management workflows that will, for example, can include contract generation and also electronic signatures. We are basically integrating all these elements under one single umbrella. Relay also would be integrated with our agentic platform. For example, a customer can build a contract review AI agent or create basically a review, approval, document generation, and signing workflows easily, all within the same environment. Smart actions will also allow Relay workflows to be embedded directly into a Box App, Box Preview, or anywhere else, bringing automation all where the work is actually happening. Box Apps is also a very exciting new innovation that we have just recently released because our vision is that customers will build thousands of apps to basically run their businesses, and these apps can work directly within the Box environment.
As Aaron mentioned, we see the integration with many applications in the enterprise, but also the need to create solutions that do not need to bring other solutions to their tech stack that could be directly built next to where the content is. Box Apps is already available to our customers as part of Enterprise Advanced, and we have some very exciting updates coming this year. Let's take a look. We have been building this unified environment with a no-code app builder that basically brings across all Box products, including Relay, Box Forms builder, the broader set of capabilities and APIs, with also metadata extraction directly available to be done from Box Apps to easily go from searching from content to setting up metadata templates to customizing the extraction process in one place.
Crucially, we are integrating our agentic platform into Box Apps, where you can imagine, for example, a sales operations person that could bring an app for automated basically deal desk approvals or your HR team that could basically build an app for writing performance reviews, all using AI directly integrated within Box. Box Apps is also another very exciting part of the platform that brings breakthroughs of AI-first technology into the ability to curate content and that sets apart the world of content portals. In traditional portals, the creation of the portal itself, the maintenance is heavy IT-dependent action. In this case, you can have a single file that could be added to different hubs at once where the data is not really replicated. What Aaron mentioned earlier, that sometimes content needs to be kept for all these different applications and needs to be copied or repeated.
Here, you can create hubs for many different uses without copying the content. Also, the curated sets of content of many different types with heterogeneous files of many different kinds could be queried at once using AI. Truly a game changer to basically do RAG and to have the ability to not only publish and create content, but apply AI to it. We are going to continue to invest in hubs and AI strategy. We are increasing the limit of files that you can have for each hub to 20,000 per hub and up to 10 million hubs per customer. With this level of scale now, virtually every enterprise could benefit from using Box Apps. We see also exciting adoption of hubs across virtually every industry.
Search and AI also will be integrated in hubs, and we will have also the ability to have users to set highly relevant information unbelievably fast. As new AI capabilities get developed in the market, we see all this acceleration of creations of models. We can bring them directly onto hubs at the same time of creating customized AI agents for a hub or the ability to use hubs for deep research. Now let's take a look at the quick video.
These critical workflows, from closing sales deals to onboarding new talent, power every business. Many of these workflows are done manually and across different tools, slowing down key decisions like whether to renew contracts and when to make job offers. With the right technology, teams across your organization can navigate these challenges correctly and on time, every time, and leave the rest to automation. The all-new Box Apps lets you empower your people to create bespoke no-code applications and dashboards powered by metadata so everyone can easily access what they need and quickly find information. For example, with a contracts app, users get a clear view of projects that need attention, like expiring agreements. See key information that's missing, like contract dates and amounts? Box AI can enrich your documents in one click. Have a contract that needs renewal? Automate it.
Watch as Box Relay seamlessly kicks off the process with the all-new Box Forms. Relay then hands the workflow off to Box Doc Gen, also brand new, and a new contract gets generated in seconds. Next, Relay routes the file to Box Sign so you get all the signatures you need. Voilà, a completed renewal lands back in your dashboard. Close those deals, fast-track your invoices, onboard that new talent, and get work flowing with AI-powered apps and automation that your teams will love. The possibilities are infinite.
Cool, right? This shows how all of these different components of the portfolio are integrating into one complete environment where we can bring apps and workflows and AI to basically create an environment to really extract intelligence from content. I'm very excited about these new capabilities and our vision. Before we move into GTM strategy that Olivia is going to cover in a moment, I want to cover two other important pillars: how we extend and integrate these capabilities with other components of the ecosystem, and how we protect and manage customers' most valuable data. As you know, Ben already shared an overview of our Box AI ecosystem, and we're bringing basically many partners to the mix. We're also bringing Box AI directly where our users collaborate on content with integrations like Box for Slack, ServiceNow, and Microsoft.
We're also using partners like VEED and LILT to bring partner AI into the content that is on Box as customers need it. Finally, we're also bringing other AI apps into Box to directly edit the content and create powerful content. For example, most recently, we integrate with Adobe Express directly within Box. In addition to this, we're also doubling down with our largest and most strategic partners like Salesforce and Microsoft. We have deep integrations with Salesforce Agentforce that is a great model for getting AI productivity while maintaining data protections. Agentforce can query Box AI directly, and basically Box AI will respond using the information and the permissions set up in Box. We basically have the ability to leverage Box governance that is very appreciated by our customers, but also the benefits of Agentforce in this scenario.
This allows enterprises to avoid data silos and basically to get the best of Box AI while maintaining control of their data in one environment. Similarly, we have already implemented Copilot via a Box Connector for Graph, so users can leverage Copilot capabilities for content on Box. This basically means that you can quickly synthesize and summarize content, but basically do it on Box and Microsoft, even in Microsoft Teams, that cuts down on data silos and unlocks AI productivity while giving enterprises flexibility and control over AI and their content. These areas are very much essential in terms of integration, but as always, basically protecting our customers' content is paramount. Basically, we always want to make sure that advanced data protection and compliance are always taken care of. Along with the benefits that AI brings, there are also additional new risks.
Intelligent content management needs a new approach that builds data protection, governance, and compliance into content workflows directly integrated instead of being a bolt-on separate thing. For that, at Box, we have taken a holistic approach to really provide data protection and compliance across the full content lifecycle. As we focus on ICM, we're making now basically security offerings stronger and more intelligent by leveraging AI into security as well. Now, classification can be the most potent tool for content security. LLMs have this extraordinary power to basically evaluate content at a much deeper and greater speed than ever before. That is why this year we'll be launching Shield AI with auto classification. It can detect, for example, a pre-released movie script versus a released script, classify files based on company policies, and apply watermarking.
In addition to that, also to defeat content sprawl and to meet compliance requirements, many organizations need an effective archive solution. Based on customer requests, we have decided to build, and now in April, we are launching Box Archive. Box Archive enables basically customers to sequester content from general circulation to keep things tidy and at the same time preserve the final state of that content. Archives can be easily searched, and they can be restored and are built directly into the same platform that our customers trust, but without limits by size or type, so customers can preserve all the content that really matters. Later in the year, we're also going to add policy-driven archives and then in the future, AI-driven archive to make archiving data as streamlined and as consistent as possible to keep basically teams very much productive.
Finally, we're also introducing data security posture management. DSPM basically is important because unclassified content can have unfettered access that could pose significant security risks to organizations, so much that Gartner believes that more than 20% of organizations are going to deploy a data security posture management solution by 2026. With DSPM, customers will be able to quickly understand the sensitivity, access levels, and permissions of content on Box and quickly take appropriate actions to basically continue to improve their security. Now, with the recent launch of so many critical new products and with so much innovation to come, we're driving the future of intelligent content management. We have built now the implementation of all-in-one content basically and an AI platform that our customers can partner with and rapidly evolve as they work into the leveraging of AI in one implemented and unified environment.
As the last slide, I wanted to quickly share our highlights on how we're using these tools at Box ourselves as an example of how companies can leverage it. Let's take a look. We're always exploring ways to basically make ourselves more productive to drive faster innovation. In the product team, we focus on how to move things as fast as possible. For that, for example, we're finding that hubs and AI and using multi-dot queries is a very powerful way to really give overviews or answer questions and reference specific data points. It's something that I'm sure you can also use. It's been great for setting, for example, onboarding and training hubs for new employees. We use it also for many different departments, and we use it also to answer questions and refer to the right documents.
Another thing that might be useful for you directly is actually the ability to do market research using multi-doc and multi-model select on groups of documents that can include also web resources and do deep research agents. We're also starting to use Box AI agents to create reports and do report generation to basically allow teams to focus on data analysis and not so much on the generation of the report, basically. I'm very excited to see this great opportunity of leveraging every one of the tools that we're building also in-house. With this, I would say that the opportunity ahead of us is quite extraordinary. I'm very excited about the strategy that we have set out to capture it.
Before I turn over to Olivia to talk about go-to-market strategy, I would like to invite you to take a quick 10-minute break and grab some coffee and resume right here. Thank you very much.
All right. Hi everyone. Let's see. Is everyone back? Yes. Okay. I'm Olivia Nottebohm. I'm the Chief Operating Officer at Box, and I'm going to be talking to you today about the go-to-market strategy. We'll do a quick look back on FY 2025 and then look forward into all that we have in store in this coming year. The go-to-market strategy for the team is about driving profitable growth in the long term. When we think about that, we actually set out on this journey last year.
For those of you who were able to join us at Financial Analyst Day last year, we had four key go-to-market strategies. The first was to expand and monetize into new markets. The second was to advance our platform opportunity. Our third was to drive more into the partner ecosystem and make it more robust to help us scale. The fourth was to drive awareness and consideration of Box as an AI-led solution. All right. How did we do? We went into three key areas. The first was AI and content, and you've heard a lot about that today. We also went into workflows. That was the e-Advance SKU that we released in the back half of Q4, and we were able to do a few dozen deals on that SKU, which was great to see the momentum. We also further expanded into ECM.
We were already able to replace certain use cases with ECM of the legacy ECM players. With e-Advance, we were able to take on even a broader surface area. We also saw many interesting use cases, and Diego alluded to those earlier. What was interesting for me as we looked across the deals that we did in Q4 was that they were both vertical-specific deals and those use cases, but also horizontal in nature. We saw things like approval workflows or contract management, these use cases that could be used really in any industry across teams. We also saw more vertical-specific use cases, like in financial services, where we saw loan origination or claims management. We were also happy to see that we were able to drive logos, new logos, and also upgrade existing customers.
Really a mixed approach here as we were going out into the market at the very end of the year, but eager to see that momentum building. All right. What did we see our customers use Box AI for? It really fell into three categories. The first was they were driving insight out of their content. Not only were they doing that in order to drive better decision-making, but they were actually leveraging that to then generate more additional content that was more relevant, saving them work, etc. Of course, they were driving these AI-powered workflows, and I'll do a couple of examples of that, and Mark will talk more about customer examples later. They were also, interestingly, building purpose-built custom agents.
What we care about as a go-to-market team is how our customers actually drive business value from those different applications of Box AI. In the content plus AI category, we saw customers like Miller Tanner were actually able to save over 800 hours of work that their teams were doing because they were extracting that metadata from the travel information in different languages, templates, and formats. They were just sending that information via API to all of the event planners across their team. A tremendous amount of time was saved just in the early implementation of it. When you take a look at workflows, there is a company called Funwell that was able to save only after 1,500 hours, actually, of time saved. This was fantastic to see.
What they basically did is they took an entire loan application process where people were submitting bank statements, etc., and people were having to receive it on their end and making decisions and passing it along this process. They were able to fully take that information with our metadata extraction tool, drive it through a series of APIs and along a workflow, and remove that work entirely. Not only did this save Funwell time, but it also increased the match rate of the loan application process. Another example that we have is one, again, in this purpose-built custom agents that I spoke about, a company called IDEX was actually able to build an agent that helped them flag early on elements that they thought in their content would actually get audited and take preemptive steps to make sure they were more secure.
The second pillar that I spoke of was expanding on our platform opportunity. Here we saw a dramatic increase in terms of the number of chargeable APIs, which is wonderful to see. This came on the backs of us releasing new APIs, but also a better UI environment, as well as a stronger sandbox. Here we saw developers able to leverage this platform and be running workflows and developing those workflows on our platform. The third part of our strategy was to extend our partner ecosystem. We did this through SIs and ISVs, and we had a number of announcements, including Slalom and Salesforce, and we had vertical ISVs that we also made announcements with. We were going to market together. Of course, the hyperscalers. We really leaned into this partner ecosystem, and we are excited to see the momentum there.
Finally, the last pillar was growing awareness and consideration. This is where we got what we called the triple crown across Forrester, Gartner, and IDC. We were awarded to be in the leader quadrant across their various categories that dealt with intelligent content management. All right. Now let's turn to FY 2026. What do we have in store? First and foremost, we want to be the leader in delivering an ICM platform to our customers. How are we going about that? It has four key elements. The first is to be the leader in content plus AI. The second is to deliver intelligent automated workflows. The third is to deliver through platform and partners, so very much what you heard about from last year and pulling it through to continue to scale.
Finally, we want to make sure that we're nurturing and advancing our install base. Let's dive into each of those. First, establishing Box as a leader of content plus AI. There are five elements to this. The first is continuing to drive on that industry leadership. The second is continuing to push through our digital channels and making sure we have even more surface area in how we get the word out. The third is continuing this in-person muscle. During COVID, there weren't as many events. I'm sure all you all feel now everything's back. We feel it too. We want to be out there even more, not only with Box-led events, but also third-party events. Industry events and partner events where we're showing up and engaging with prospects and customers.
Of course, we're working on the customer stories and codifying that business value where we see customers capturing real, real dollars, hours saved, opportunity they see, faster innovation, making sure we're capturing all of that as reference customers so that other customers can come on board. Of course, ongoing marketing tactics like influencer marketing. All right. You heard about a number of use cases. We saw those early use cases in Q4. That was from a few dozen deals we did. The interesting part is we continue to see additional use cases. Now we can see the pipe not only for Q1, but stretching out into the year. We're seeing additional use cases in financial services, and we're also seeing additional industries come in. Now we're seeing examples in healthcare like patient onboarding.
We're seeing examples in construction like project management or field operations automation. We are also seeing additional horizontal use cases, so client onboarding. That can happen across a number of different industries. We are incredibly excited to see that our customers and prospects are continuing to think of what they can do on our platform, and we see this momentum building. We continue to push forward on platform. Obviously, this has been a theme throughout. We've published our API catalog. We brought APIs for AI into the market at the end of Q4. We now have AI units. We're seeing momentum building there. Obviously, the DocGen API is super interesting to our customers as we're talking to them about creating these workflows. We could not be more excited to be leaning in and driving the awareness with the AI developer community.
As our partners, whether they're SIs or ISVs, are building on our platform, we are making sure to be bringing these latest frameworks to them, but also the latest LLMs and making sure we're always at the cutting edge of AI. I spoke about last year about how we were pushing into the partner ecosystem, and I said, "Really, our focus is going to be on SIs and ISVs." We are continuing to do that. We've inked a number of partnerships this past year. We will continue to do that this coming year and really leaning into the set of partners to make sure we go deep. These partners make money on the back of Box. We know that the winning formula for the partner ecosystem is that these partners are out there.
They're able to deliver services dollars on the back of our solution, and they're excited ultimately to be bringing us more deals in the future. This is how we scale. Last year, we think of it as year zero. That was early days because now we were in the market with a solution that had real services dollars to offer at a scale that partners would be interested in. Now we're looking at a year where it's all about deals, just winning deals together. It doesn't matter where we source the deals from, but we want to be out there winning deals with partners. Again, the reason for this is, first, we believe we're going to deliver tremendous value to our customers, but also because we know that if we can make our partners successful, we too will be successful.
Of course, in the following years, we very much hope, DataBank, that you will be bringing customers to us, for those partners in the audience. We can actually be scaling this with this exchange of, "Okay, we bring some deals, partners bring some deals," and that's how we 10x it together. Finally, a number of questions have come over the transom about where are we putting these investment dollars this year. These investment dollars are going into three key areas. The first is content and AI. I spoke about that marketing push to drive the awareness, to drive the consideration, because what we see is that when we're out there talking to customers and prospects, when they hear about intelligent content workflows, they are so excited. We just want to make sure to be getting the word out even more broadly.
The second bucket is sales. We have actually invested in additional sales capacity. We also stood up a vertical sales team this year. We already have a public sector one. We are now live with a financial services team that is verticalized as well. The final one is partner. We continue to invest in partner. We put in dollars in terms of co-marketing and delivering POCs together, but we also are hiring more folks to be right there facing off with the partners and going into these deals and having those conversations. Finally, we talk to our customers all day long about Box AI. The natural question is, how does Box use Box AI? For go-to-market, we use it across all of our go-to-market teams. In the first instance, for sales, we use it, and Ben mentioned this earlier, this ability to do RFPs much faster.
Our sales team also get RFPs. We use that same metadata extraction tool to understand what the ask is and then to generate what the proposal would be. That saves a tremendous amount of time for our sellers. From a marketing perspective, we put all that marketing collateral in a hub, and we are able to not only ask for insights, but actually generate proposed content for future campaigns, future copy. That saves our marketing team a tremendous amount of time. On the customer success side, we, of course, have hubs built for customers. We know what the customers' care abouts are and what might be some of the questions they're asking. Those CSMs are going into those hubs and getting answers to the question that the customers are asking them. Finally, in support, we have something called Support Sensei.
This is also a hub where our agents are going in and asking the questions coming in over the tickets, and that hub is then recommending what the resolution would be. This not only drives down our time to resolution, but it also means that we can onboard agents over 30% faster when they're bringing them on and able to use Box AI in this hub. Really exciting. I can't wait to tell you even more next time we get together. Thank you for your time today, and I'll turn it over to Mark.
All right. Thanks, Olivia. All right. We'll just get it started with, we'll address this thing. I had hand surgery last week, and I fell on a rock, and the rock fared much better than I did. Unlike the rest of my coworkers, I'll be doing this presentation one-handed. Keep seated. No applause needed. All right. That's my hand, and let's get on with our sales strategy here for FY 2026. To start with, as you know, we're a global company. We sell Box in many countries all around the world, but these are the seven countries where we have really teams on the ground. In our subscription business model, we need a team of 10-ish Boxers in order to support a customer in a region. We need salespeople, solutions engineers, managers, marketing, legal, and on and on and on.
This is where we have teams on the ground, and we're in these markets because they're the largest markets on Earth for enterprise software. Of course, through digital and light touch approaches, we serve other corners of the world as well. If you think about the evolution of our sales strategy, we've really been evolving our sales strategy over these last 10+ years as the product and as the company has evolved. If you think about our founding sort of sales motion, when we are doing file, sync, and share, that was really a motion that was direct sales complemented with solutions engineers. Then you move into the second chapter of our growth as a company where we really start moving into these more external collab and securely managing content. We start integrating to more systems.
We still have that direct sales motion with sales and solutions engineers, but we start our Box Consulting practice that can help with these more complex implementations. As we move into this new world that's aided by Box AI and intelligent content management, much more sophisticated use cases, we add in additional solutions engineering support with some specialists and overlay functions. Oftentimes, these are solutions engineers that have been with us for quite some time, but in a prior chapter of their career, they might have worked in enterprise content management for one of the big-name ECM providers or for a large consulting firm. We end up with some specialization around these ECM-oriented workloads. Of course, as Olivia just covered, we're having a huge expansion of our partner ecosystem, working with global systems integrators and regional systems integrators.
We'll talk to DataBank today so we can handle more of these increasingly complex sorts of use cases. All right. Three topic areas today for FY 2026 that we'll do a little double-click on would be Enterprise Advanced, what we're doing in verticals and industries, and of course, what we're doing with partners. To talk about Enterprise Advanced, you've already heard about the product from Diego and from Aaron and from Ben. Let's just talk about some customer examples, and I have a couple of really interesting ones to share with you. The first one is with the Texas DMV. This is the largest department of motor vehicles in our union. They had a problem in that they had a legacy ECM tool that was sort of nearing end of life and didn't have a lot of innovation coming onto that platform.
Because this is such a massive Department of Motor Vehicles, imagine all of the unstructured content that they need to manage with vehicle registrations and driver's licenses, lots and lots of documentation. It is a set of documentation that just goes up and to the right forever. They had huge maintenance costs on this legacy platform without a good innovation path ahead of it, and they kept having to throw more money at it for increasing amounts of storage. They are doing a massive digital transformation initiative that includes Box as a foundational component of it, Salesforce, and other tooling as well to modernize the Department of Motor Vehicles and create a much better client experience for those of us that go into DMVs and also a much better experience for the employees and dramatically lowering their cost of service.
They will be soon from now, if you live in Texas, your vehicle's title will be stored in Box, and the key data from that title will be extracted using our metadata extraction capabilities and put into a system of record for managing that whole process. That is an exciting one. Another one firm that many in the room will be very familiar with is AlTi Tiedemann. This is a wealth management firm that works with ultra-high-net-worth customers. As you can imagine, in all of your business, you are sitting on mountains and mountains of unstructured content, lots and lots of files. The interesting thing about the files that they are dealing with is that lots of financial disclosure documents, while they are technically unstructured, there is a certain set of the content that is inside of that data, inside of those files that is structured.
When you look at an annual report, you look at a 10-K, they're structured to that unstructured content. They also deal with a lot of alternative investments. As you know, alternative investments don't have the same sort of structure to the financials. If you're looking at a real estate investment or that sort of thing, those documents come in all shapes and sizes, and they don't come with the same sort of structure. What they are doing is using Box actually to create a number of hubs that wealth advisors can put a whole bunch of content in there, both structured and unstructured data, lots of alternative investments in there. Using multi-doc queries with Box AI in that hub, they can ask questions of that. They can allow their wealth advisors to make investment recommendations much quicker.
They can also create client-facing hubs where all the clients' documents can be in that hub, and they can do multi-doc queries against those documents using Box AI. Pretty exciting stuff. From an industry standpoint, as we've shared before, we do very, very well in industries that have large volumes of highly sensitive content. Especially when they need to share that content internally and externally, we end up being a very, very good fit. We verticalized our public sector business a few years ago. This year, we're moving into and started verticalizing in financial services, which has historically been our number one industry. As you can imagine, in media and entertainment, most of the TV shows and movies you've seen over the last few years, at one stage in the creative process, were managed on our content cloud.
In public sector, we're doing very, very well in child protective services, in the Department of Justice, all sorts of agencies in our government where there's lots of sensitive content that needs to move between agencies or move between citizens and the agency itself. Of course, in life sciences, as you've heard before, the next generation of drugs are in the clinical trials process, for example, very much managed on Box. Okay. That brings us to partners. I think that Olivia covered fairly well our partner strategy. Just to start with, as you know, we have resellers, we have ISV partners, and we have systems integrator partners. We're making big investments across all three. There's great synergies between all three.
Take, for example, our partnership with Salesforce, where they have industry clouds that align very well to our strategy, for example, life sciences cloud and our financial services cloud. We can integrate to those clouds and have very compelling use cases to help secure sensitive content in those interactions. In addition, we can then partner with Salesforce ecosystem partners like Slalom that's now a Box partner. You have this sort of beautiful thing where your industry strategy, your product strategy, your partner strategy all fit together nicely. We are moving into this new world where we are taking on ECM workloads. That opens up a whole new opportunity for us to partner with a number of firms that have been in that ECM world for sometimes two and three decades. Lastly, AI plus content workloads are sort of endless, right?
From Aaron's presentation, you saw that the average enterprise has 90% of their data is unstructured. With metadata extraction and workflow and Box AI, we have the ability to really magically go into these mountains of content that are unstructured and give them structure. Once that data has structure, then the use cases are somewhat limitless. There is no better way to sort of talk about this than with one of our partners. I would like to invite Matt Charlson from DataBank onto the stage. He is CEO and President of DataBank. You saw the announcement today of our new partnership, Matt?
Yes.
Welcome to the stage. We'll do a little left-handed foot twist prompt.
Thanks for having us.
Thanks for joining.
Absolutely. Happy to be here.
Maybe if you can introduce yourself and tell the audience a little bit about DataBank.
Okay. Yeah. Like Mark said, I'm Matt Charlson, the CEO of DataBank. I've been in that role for about six years, but I've been in this content-centric industry for 25 years. At DataBank, our strategy is all around helping our customers manage unstructured data, right? That's generation of unstructured data. It's putting it through workflows, managing, destroying, all of the aspects of unstructured data. We're super excited to be here and super excited with a new partnership with Box.
That's great. Thanks for coming. Why is now the right time for our two organizations to partner together from your perspective?
As I mentioned, our vision for the last year or two has been, how do we unlock the value? You've heard a lot today about data inside of content being trapped, right? You get to a certain point where you can manage it, but it's trapped, right? In our vision, the last couple of years has been, how do we get better at that? We can manage content very well. We can manage content. I think we're extremely proficient at that. How do you get those insights unlocked in there? I think now is the right time because tools like companies like Box are developing. They're putting emphasis on getting deeper into those insights and extracting value, extracting things from the records like using IDP, using generative AI. That's going to unlock a huge potential in the market, I think. I know it is.
When you think about what we're seeing with AI and broadly this category of intelligent content management, how do you think this is impacting traditional ways that content's been managed in the enterprise?
Let me take the workflow example as one because I think that's one that's near and dear to everyone's hearts here. What I've learned in my time managing teams, constructing these workflows is there's a basic anatomy of any kind of workflow, right? You've got to find the right data, get it to the right person at the right time so they can do a task or make a decision. That's a manual process almost always, always has been. It doesn't really matter what industry it is. That's the basic premise of the workflows that people like us are out there building. Someone's got to, once all that gets routed to the right person, that's all got to get read, right? I've got to read 10 pages. I've got to read 100 pages.
I've got to read all this content so I can do my task, make my decision, whatever my step in that process is, right? The generative AI tools that we're hearing about now are going to help the user read that. If I'm reading through 100 pages of information, I'm bound to miss something. There's a nuance in there that I'm going to use. I'm going to make the wrong decision. I may not have any insight into what I'm looking at. Every case, every transaction is almost a new thing, especially if I'm new in that process or maybe new with that organization. I think it's really important that we leverage these GenAI tools to help summarize, help provide insights, maybe even make recommendations.
When you take that to the next step, when you start thinking agentic, we're going to be able to automate those processes and get more accurate to make those decisions. We're not going to need the humans to do all of those steps anymore.
When you think about the Box partnership, is this an opportunity? How does it change your go-to-market strategy? Are we expanding your addressable market? How do things change for you?
Yeah, I think, again, you hear all the studies. The study we heard today was 90% of any enterprise, their information is unstructured, right? That means right now it's trapped and it's locked, and there's not a lot of usefulness to it other than we've got it. We've got it managed. I think it's going to untap a huge potential for companies like us in terms of being able to go extract that information, provide insights, feed a data domain that is structured, that is missing a lot of key components. I think it's going to open up. Think about ECM. It's been a stable market. It's been kind of a flat market for the last few years. I think this opens up a ton of potential for us.
That's great. Lastly, a big part of your business is this business process outsourcing part of this business where you go and pick up documents and scan them. How does that part of your business really change with the power of AI and automation?
We have seen an influx the last year. We have seen an influx in our pipeline of organizations looking at their massive paper repositories, right? In the past, hard to justify the investment that a company might need to make to do that. Now they are looking at it, "Hey, I can get value this. It's not valuable enough for me just to scan it and archive it and then have it there for retention purposes." What is happening right now is I am able to say, "Hey, look, I can take that data and I can use that to feed an LLM. I can use that to maybe understand my go forward risk by looking at my history." There just has not been tools to do that in the past. We have the massive capabilities to scan and index that information, but no one is in there mining it for future insights.
We're already seeing it happen today. It's been a pretty powerful thing.
That's great. Matt, we're grateful for the partnership. We're excited about what we can do together. Thank you for joining us.
Happy to be here.
Thank you.
Appreciate it.
Thank you. Thank you. All right. With that, I will hand over to Dylan Smith, Co-Founder and CFO.
Awesome. Thanks, Mark, and thank you, Matt. As mentioned, I'm Dylan Smith, Box's CFO and Co-founder. You've now heard why Box is so well positioned to attack our expanding market opportunity and how we're extending our ICM platform's capabilities as well as augmenting our go-to-market motions to capitalize on this opportunity. Now, I'm going to close this out by discussing how all of this translates into our path to achieve double-digit profitable growth and a top-tier financial profile. I'll start with an overview of how our financial profile and key metrics have evolved over the past couple of years. Then I'll walk through how we expect the core growth drivers of our business to evolve going forward and why we're confident that we're firmly on the path to achieve and sustain double-digit revenue growth.
I'll end by sharing how everything you've heard today ladders up to our overall financial strategy to deliver shareholder value as we steadily execute toward our long-term target model. This past year in FY 2025, we navigated a challenging macroeconomic environm ent as well as currency headwinds to deliver strong financial results exceeding the high end of our guidance across all of our metrics that we guide to from top to bottom. I will dive into how the foundation that we've been building that you've been hearing about, whether it's our financial foundation, product foundation, and go-to-market foundation, all combine to drive the momentum that we're seeing in the business and set us up to capture our massive market opportunity as we embark on Box's next chapter.
Diving into the numbers, on the left of the slide, you can see the momentum that we're building in last year's RPO growth. Short-term RPO was up 6% year- on- year, 7% in constant currency, which tends to be a pretty good leading indicator of our forward revenue growth. Long-term RPO was up more than 20% on the year, and that was driven by strong early renewal volumes and longer and longer contract durations with our customers, which provides us with better multi-year revenue visibility. On the right, you can see our billings, which for the year were up 6% in constant currency and up 7% in constant currency in Q4, which was several points ahead of our expectations, again driven by the strong momentum that we saw from Enterprise Advanced out of the gate.
These results demonstrate both the demand that we're seeing for our higher value offerings and that customers are increasingly viewing Box as a critical partner in their long-term IT and AI strategies. We've laid the foundation not just to grow, but to grow profitably. As a reminder, when you think about our gross margin, about 18 months ago, we completed a multi-year public cloud migration that is already yielding significant operational benefits. We're now running our infrastructure in a more scalable way, a more secure way, and that allows us to focus more of our engineering resources on delivering product innovation to customers rather than managing infrastructure. This infrastructure is also having and creating significant financial benefits, which you can see here in FY 2025's gross margin.
Gross margin improved by a full 400 basis points and came in at 81%, a full point ahead of our expectations entering this past year. That is the biggest driver of our operating margin outcome for last year, which was up more than 400 basis points in constant currency, about 300 basis points on an as-reported basis, and constant currency operating income grew by nearly 30% year- over- year. Significant improvements to our underlying profitability while at the same time allowing us to invest in standing up a lot of the critical growth initiatives that we have been talking about today. Those investments also allow us to further grow and diversify our customer and revenue base.
When you combine that diversified base with longer contract durations, our 97% recurring revenue model, a best-in-class churn rate, all those things make our business more predictable and more resilient to headwinds that might impact any specific geographies, industries, or segments. Kind of diving into some of these numbers, you can see that our international business continues to perform well. That now represents 36% of total revenue. That is up a couple of points year on year and driven by continued strength in Japan. From an industry point of view, we have a very, very horizontal platform where our top three industries make up a little less than 40% of our overall business. Again, because just how applicable all of what Box's platform provides is across a variety of companies. Finally, we also serve and add value for companies of all sizes.
Over the past several years, we've seen a gradual increase in the percentage of our revenue coming from enterprise customers, which we define as customers or companies with at least 2,000 employees. A few years ago, that was about 60% of the business and continues to tick up as we scale. Now you can see that as more of our customers have moved to high-value suites over the past several years, our underlying customer economics have significantly improved and have remained strong this past year with the backdrop of a challenging macroeconomic environment. In FY 2025, and driven by the continued momentum that we're seeing with suites across our customer base, we saw strong growth in our $100,000+ customer population, up a little bit ahead, a couple of points ahead of our overall revenue growth, given where the value proposition is resonating the most.
This momentum has also been the biggest driver of the price per seat improvements that we've continued to drive at an average clip of about 4% per year over the past five years. That'll continue to be a big growth driver for us going forward. Our full churn rate remains very strong and stable at a best-in-class 3% annualized, which is an improvement from where we were just a few years ago when our full churn rate was about 5% annualized. We expect our customer economics to continue to improve as more customers realize the value of our ICM platform. With that, I'll delve into each of these categories in just a bit. Turning and starting with large customer growth, here you can see we segment that by 100,000+ , 500,000+ and million-dollar plus customers.
Over the past year, all three categories of those customers grew faster than our revenue growth. These customers are so important, both because they contribute about 2/3 of our total revenue and because these customers with larger deployments are set to benefit the most from our newer capabilities, including AI. As we continue to improve monetization and improve the pricing of our platform beyond our core seat model, the population of customers who can ultimately contribute to each of these categories is also expanding, which is really exciting for us. We're already seeing a lot of even relatively small SMBs becoming six-figure customers for us because of the extension of our various capabilities. I'd also note that with our land and expand model, that often happens over time.
If you look on the far right, just over $150 million+ customers did not get there immediately. In fact, 85% of those customers actually started as Box customers with annual spend of less than a million dollars. As they adopted the additional products and use cases that we have powered over time, they eventually surpassed that clip to where they are today. Turning to suites, as we have discussed, selling suites and Enterprise Plus in particular historically has really become our de facto sales motion. We have been really, really thrilled by how customers have been adopting and embracing suites to leverage the full value of Box's platform. Suites customers now represent 60% of our revenue. That is up from 55% a year ago.
We're also applying all of the learnings that we've made through driving Enterprise Plus adoption and applying that to how we're rolling out and driving Enterprise Advanced. That's already seeing early success as we've talked about. I would note that while suites as a percentage of revenue is an important indicator of just what portion of our overall business is leveraging the full value of Box's platform, it's not necessarily capturing the impact or volume of suites bookings in a given year. Just because due to the nature of the metric, that percentage tends to move a lot more when a customer moves into suites for the first time versus when an existing customer might have a significant upsell and remain a suites customer, the metric doesn't move.
For example, this past year, you can see we added five points to that suites penetration rate versus nine points in the year prior in FY 2024. Actually, in FY 2025, suites bookings were up about 10% year on year and growing certainly faster than they were in the FY 2024 time period. This chart almost understates the momentum that we've been seeing in suites over the past year, although it is a very important metric for us. Turning to the other ways that our customer economics are evolving, here you're looking at our net retention rate or effectively what is the growth that we've been driving over the past 12 months from customers that we had a year ago. As a reminder, our net retention rate has three components.
There's seat growth and pricing improvements minus the churn that we see from our customers altogether lead up to that net retention rate. Over this past year, we achieved our goal of improving our net retention rate to 102%. In the coming year, we expect our net retention rate to tick up further to 103%, driven primarily by pricing improvements. Longer term, we expect the growth in our install base that our net retention rate represents to be an important driver of our return to double-digit overall growth. We expect that net retention rate to be in the 105%-110% range longer term.
The trends that I just discussed in our model and our customer economics really provide the foundation to deliver that durable double-digit revenue growth that we've been talking about and continued margin expansion in FY 2026 and beyond. Next, I'm going to walk through how Enterprise Advanced and AI support and drive some of our key growth initiatives, which is a big reason that we're so confident that we are firmly on the path to deliver that accelerated revenue growth. We've, of course, talked about Enterprise Advanced and Box AI quite a bit today, and for good reason. I'm not going to be an exception to that just because they are such fundamental catalysts to accelerating our growth rate in the coming years, with just a few of the key growth drivers in way that can show up in our business outlined here.
First, we're confident that based on everything we're already seeing in these deals and within our customer base today, that will allow us to further improve the impact of pricing on our overall growth rate to 5% or more annually, as well as should help us improve kind of the seat growth trends that we've been seeing over time, which I'll dive into in just a bit. As Olivia and Mark shared, Enterprise Advanced's capabilities create a much larger opportunity for us to work with partners. We expect the investments that we've been making in our partner ecosystem to grow this business at a roughly 20% CAGR over the next several years.
We also expect Enterprise Advanced, and as we talked about a bit earlier in the day, to create a much larger platform monetization opportunity for us beyond our current primarily seat-based model, which also allows us to disrupt a much larger part of the legacy enterprise content management market. Now I'll discuss how we are approaching all of these key growth initiatives and how we expect these opportunities to develop as we march toward our long-term target model. As we've discussed previously, we typically tend to land customers by licensing Box for a portion of their employees with the opportunity for seat expansion over time. Currently, we have a roughly 7x seat expansion opportunity within our install base today and a much larger seat growth opportunity if you look at the total applicable seats in the markets that we currently serve today.
While we expect EA to have a larger Enterprise Advanced, they have a larger impact on our growth through pricing improvements. We do expect that over time, Enterprise Advanced adoption should create tailwinds to seat growth as well. For example, our enhanced workflow capabilities power use cases that extend across a bunch of different departments. Our newer capabilities can also replace spend in categories such as data loss prevention for departments that might not be licensed for Box today. We expect that nearer term and probably bigger impact to come from pricing. We really expect these newer capabilities to have a positive impact on our financial model from top to bottom.
As Aaron discussed, and while it's hard to predict exactly how AI is going to change the enterprise software landscape over time, it clearly has the potential to create enormous customer value and to transform software business models. For example, and again, going back and building on what Aaron mentioned, as AI capabilities improve, AI agents that connect different systems will proliferate. These aren't seats or representing a person in a traditional sense, but as Aaron noted, they are really sitting alongside knowledge workers and are incredibly valuable for companies. Those create an exciting business model opportunity for Box. Turning back to our more traditional growth drivers, we're increasingly confident, as I noted, in Enterprise Advanced ability to improve our growth rate from price uplifts.
As you can see here, we have a multi-year track record of steadily driving pricing improvements at a clip of about 4% per year on average over the past several years. Enterprise Advanced extends our value proposition more than any other launch that we've had in the company's history. That's clearly already resonating with our customers. As our customers continue to adopt Enterprise Advanced and our newer AI capabilities, and as we continue to achieve that 20%-40% pricing uplift over Enterprise Plus that we've already been seeing today, that will increase our pricing compounded annual growth rate to 5%+ going forward. Going back to our partner business and building on what we were discussing earlier, that business has been an important contributor to our growth historically, and it currently represents about 20% of our overall revenue outside of Japan.
Japan, just due to the nature of the market, is a business that is sold effectively 100% co-sold with partners. We exclude that here. Within that 20%, the majority of our current partner revenue is coming from traditional resellers. Going forward, we expect to see a much larger opportunity to partner with systems integrators and ISVs, including marketplaces. Once again, Enterprise Advanced and AI are key drivers of this larger opportunity for us. Companies are recognizing that how they manage their unstructured data profoundly impacts their ability to leverage and get the benefits from AI. Now that Box has the capabilities to disrupt a much larger portion of the legacy ECM market, we are really well positioned to be the technology partner of choice as these SIs are going in and working with their clients to help navigate the shift.
We heard earlier from Matt at DataBank around how they're viewing Box and the overall market opportunity. That would be a great example alongside other partners like Slalom that would show up in that top light blue bucket with SIs. Going to the middle bucket with ISVs and marketplaces, that is really building on the kind of types of relationships and integrations that we've already had and been focused on over the past several years, but really extending just how much value we can add to the ecosystem and then capture and return. For example, we're already gaining traction with leading ISVs that have been announced like Salesforce through Agentforce or Guidewire, who are seeking to extend their customer value proposition by integrating content and content insights into what they do.
We have also recently announced kind of new marketplace relationships with both GCP and AWS. Resellers, there on the bottom, you can see will remain an important part of our partner ecosystem. We expect that business to continue growing. We just do not necessarily expect it to drive a meaningful improvement in our overall growth rate. Our success in Japan has certainly proven how important having the right partners and strong relationships with those partners can be for Box. History proves just how influential the partner ecosystem can be as software companies scale. We are super excited about this opportunity and all the different ways and implications of executing against it. Turning to our platform business, that is another example where we have already started to see momentum, as Olivia had highlighted, with a much larger potential going forward.
Currently, about 5% of our recurring revenue comes from platform customers, where the value we're delivering to those customers and how we monetize that value is driven by consumption volume rather than by seat volume. Examples historically would include powering client-facing portals or a loan origination process or a particularly heavy Box Sign use case. With Enterprise Advanced product capabilities, that really significantly expands what customers can do with our platform. That includes metadata extraction, document generation, and the ability to build custom AI applications and agents with much more to come. Just last month, we announced AI units to help our customers benefit from the countless ways that they can leverage our AI capabilities at the same time simplifying kind of API usage and tracking for those customers.
We expect our platform business in the coming years to grow at a roughly 30% CAGR and to contribute at least 10% of our total revenue at the end of that time, up from about 5% today. As I mentioned, it has been really exciting to see how customers have responded to our newest offerings. This slide represents a customer, a large customer of ours who has been with us for more than a decade, signing up back in FY 2014, adopting all of our offerings over time. What is really interesting is that this is not necessarily a customer that has traditionally been an early adopter of Box. They moved into enterprise back in FY 2018 when we already had a bunch of other more premium products available. Enterprise Plus, they finally made the transition there several years after that was available.
With Enterprise Advanced, literally within weeks of being able to do so, they were sold on the value proposition and went all in with Box and Enterprise Advanced as a way to automate a bunch of really important workflows, especially around some of their critical financial processes. That just, I think, is an exciting just one example, but a clear demonstration of just how much that value proposition is resonating with our customers. You would also see that led to the largest kind of pricing increase they have ever had in moving from plan to plan, both in percentage and in dollar terms.
Now that we've covered kind of how we're building the foundation to capitalize on our opportunity and how our product and go-to-market strategies and initiatives provide a clear path to deliver double-digit revenue growth, I'm going to talk about how, as we execute against this opportunity in the year ahead, we're confident that the resulting improvements in our financial profile combined with our disciplined capital allocation strategy set us up to deliver sustained and significant long-term shareholder value. We'll start by kind of recapping and bridging kind of what we're seeing in the core business today and how the momentum that we're seeing with our newer growth initiatives ladder up to why we're so confident in that ability to drive double-digit growth.
As reminded, these core drivers of growth are not going away, and we still have many, many years of headroom within those growth opportunities, whether that's Enterprise Plus or Japan or our reseller partners. Today, we focused on the parts of the business that we expect to be accretive to our current growth rate in the years to come. I'll add on those going from left to right. As mentioned, I think the biggest impact that we're going to see is from our existing customers. Those currently contribute about 70%-75% of our new bookings in any given year. We expect that ratio to generally hold, even as so many things are evolving about our business.
As we think that customers who are already kind of managing their content on Box have already bought into the value proposition, are the most likely to adopt and get even more value with our newer offerings. Next, we do expect to deliver another one to two points of incremental growth coming from bringing on a higher volume of new customers and larger customers as well. Again, really driven by our expanded capabilities, as well as the reach and impact from the strategic partners that we've talked about, as well as from growth from a new customer point of view, growth from emerging international markets such as an under-penetrated market such as EMEA, Canada, or Australia.
Lastly, going back to the platform expansion opportunity, as that rapidly evolving business goes from roughly 5% of revenue today to 10%+ three to five years from now, we expect that would add another one to two points to our overall revenue growth rate. Turning down to a similar view of our kind of bottom line expectations over the next three to five years, you can see we do expect to drive significant operating margin improvement in the coming years as well, roughly 100 to 200 basis points per year on average. Once again, we'll kind of outline some of these, which we'll then dive into in a bit going from left to right.
First, we're doubling down on our location strategy given the success we're seeing in terms of the talent and efficiencies that we've been able to kind of generate and benefit from those locations. While we've seen a step function increase in our gross margin over the past few years that we talked about, we're not done yet as the team continues to focus on and discover new ways to optimize our cost to serve now, which I think will add another one to two points to our overall gross margins, which would also benefit from the tailwinds that we're seeing from pricing improvements.
Of course, on operating efficiency, a lot we're doing to just optimize and drive efficiencies and productivity across the organization, whether that is just being very focused and data-driven around the go-to-market investments that we're making and tuning those over time to the leverage that we think we can drive by kind of building out and investing in our partnership opportunity to the way that we're running the business internally. As you've started to hear today, and you'll hear from me in a bit how AI is fundamentally transforming the way that we're thinking about how we manage different processes and kind of leverage technology internally. All of those things ladder up and combined with, of course, our continued focus on cost discipline to drive that operating margin in the mid to high 30s three to five years from now.
Double-clicking into the workforce and location strategy, over the past few years, that's been a big focus for us. We've been really excited about the way that this has been impacting our business, especially the progress we've made in Poland, which is primarily an engineering center of excellence. We've made significant progress that we now have 20% of our total employees in lower-cost locations, most of those in Poland. That's an improvement from about 14% a year ago. That shift has been more pronounced and most pronounced in our R&D organization with nearly half of those employees working from lower-cost locations, which has been one of the reasons we've been able to accelerate our product innovation without seeing a commensurate increase in costs. In the coming years, we will continue to invest in Poland.
At the same time, in the coming years, we'll be looking for other opportunities to drive similar efficiencies, for example, a location in the Americas that could directly serve customers or internal stakeholders who are located in the U.S. As you saw earlier, this focused effort we expect is already having a huge impact on our business. We expect that to be the single biggest discrete driver of the leverage that we'll generate as we scale toward our long-term model. Turning to our capital allocation strategy, which we also remain focused on, we think about that as a really important way to deliver shareholder value. That's kind of the lens that we view our overall capital allocation strategy. It all starts with the free cash flow generation. Last year, we delivered more than $300 million in free cash flow.
We expect that in the years ahead to grow in the mid-teens percentage range. That is what makes all the rest of this possible. A big part, and where we have been very successful, is with very strategic kind of focused, targeted acquisitions that we have been able to very quickly then integrate into our platform natively. Whether that is how we introduce Box Sign or more recently integrating our Cruise acquisition into a lot of the core functionality of Enterprise Advanced, like workflow and metadata offerings, to the even more recent acquisition this past summer of Alphamoon to bolster our intelligent document processing capabilities. These are all very efficient and very successful ways that we have been able to bolster our internal organic development in a really, really kind of rapid and thoughtful way.
Of course, share repurchases will continue to be a big focus of our capital allocation strategy. That is where we expect the majority of our free cash flow generation to go. We recently expanded our buyback program on our most recent earnings call by $150 million to continue those efforts. That kind of consistent share repurchase program, combined with our focus and our ability to reduce our equity burn rate over the past few years, has allowed us to steadily reduce our total shares outstanding over the past few years. We will just briefly give some backdrop and color into what is fueling that capital allocation strategy. Again, with the cash flow that we have generated, we have been able to return $650 million to shareholders over the past three years.
That represents about 80% of the free cash flow we've generated over that period. We expect to deliver a similar approach and strategy going forward. In terms of that free cash flow, again, expect that to follow a similar trend as our operating margin expansion. For those to be pretty closely correlated, although free cash flow margin should be slightly higher than operating margin just because billings and collections tend to lead revenue growth. It really is this strong free cash flow and our confidence as we both grow profitably and expand our margins that allows us to put up significant free cash flow generation growth in the years ahead. Now, just turning to kind of cover how we, with my organizations, are becoming a part of Box's transformation to an AI-first enterprise.
We'll start with kind of our legal team, where the cool thing about Box is that we can now both capture insights as well as kind of kick off and automatically kick off new workflows and actions, not only based on the data that's sitting in the content on Box, such as legal contracts, but also at the metadata level. There is a really, really rich set of information that opens up really, really powerful use cases that you traditionally need a pretty robust and expensive contract lifecycle management system to be able to address. Turning to the finance side, some of my organizations have engineers, whether that's data engineers, security engineers, which have similar use cases and are delivering similar value, as what Diego had mentioned earlier.
At the same time, across the organization, as a service organization, really Box Hubs is the lifeblood of how we are kind of getting information and responses to Boxers faster than ever before. We're not yet maxed out on our 10 million Hubs entitlement, but we aspire to get there someday, as that's been an extremely powerful way for customers or for employees to be able to get answers to the questions they'd have that they'd normally be asking of a human or wouldn't know where to find in seconds, which is better than even a service-oriented support team's time would be. I'd say that applies also to a lot of the HR use cases, really having Hubs power things.
At the same time, a lot of different kind of content-centric workflows, business processes, like in the HR example, that includes open enrollment, new hire onboarding, and things like that. On security, mentioned not just the engineering side of the house, but there's also similar use cases to what Olivia had mentioned in terms of RFPs. They do not just come into sales. Security compliance get a lot of those as well from our customers. The ability to answer those crisply, accurately in a fraction of the time allows us to do much, much more in terms of improving our internal security and compliance.
As a result of some of the tools that we're building using AI, as well as the integrations that we have with other third-party tools, we've been able to dramatically improve both the accuracy and timeliness of our kind of anomaly detection, incident response times, as well as kind of figure out the path forward and resolve those faster than ever before as well. Now we'll walk you through, and we've covered a lot of the components, but I want to put it all together and walk through our long-term target model from top to bottom. Starting with revenue growth plus free cash flow margin, we expect to deliver a combined outcome in the 45%-50% range three to five years from now, which is a 10+ point improvement from the results that we delivered this past year in FY 2025.
Turning to the revenue growth, again, a core part of that is our delivering and returning to double-digit revenue growth in that 10%-15% range. That is a result of all of the kind of existing and new growth initiatives that you have been hearing about from the team today. On the bottom line, maybe flipping to the very bottom, you can see that that kind of recaps the operating margin improvements and leverage that we expect to drive. In recent years, as you have seen, gross margin has been doing a lot of the heavy lifting here. Going forward and over the next three to five years, we expect to see a pretty balanced kind of breakdown of how we generate leverage in the business, with the majority of that coming from OpEx rather than gross margin.
The team has already covered a lot of the ways that we're driving efficiencies in our respective organizations, as well as the investments that we're making in a lot of the high ROI efforts on the go-to-market side. That, our lower-cost workforce and location strategy, and the way that we are kind of improving how we run all of our processes internally and leveraging AI are really what ladder up to the type of improvements that we're driving across the entire business. In summary, I would say that Box is well positioned to capture a growing portion of the massive market opportunity that's in front of us. As we execute against that opportunity, we are firmly committed to driving meaningful improvements in our financial profile from top to bottom. You've heard a lot about our market opportunity, our product, our go-to-market strategy.
You address that, and then you with me, how we plan to strengthen our model and return to double-digit growth as we deliver that free cash flow, revenue growth plus free cash flow margin in the 45%-50% range. I would say that if you have been following Box for a while, you know that I don't claim that this is the most exciting year ever for Box on an annual basis. Twenty years into our journey, I would say that about FY 2026. We are in the early innings of a once-in-a-generation AI opportunity, and that will fundamentally change how customers work with their data and the importance of unstructured data and building intelligence into content and content workflows. That is going to be critical for companies to be able to benefit from this transformation.
That creates a massive opportunity for Box that we've been excited to share with you all today and that we're going to be executing against in the coming year. With that, I think we will take a couple of minutes break, especially for those on the webcast, as we set up the stage and as I welcome my colleagues back up to the stage for Q&A.
We'll kick it off first with a question that we received online from an investor. Today at GTC and Jensen's keynote, he specifically mentioned you, Aaron. Can you give us some more context on this? I swear Aaron did not plant that question.
Yeah, Jensen has his main GTC keynote today at NVIDIA. There was a series of announcements around, obviously, kind of latest GPU infrastructure. They also released a new tuned Llama model called Nemotron. This is a model, and I'll have maybe Ben kind of build on this in a second, but it's a new model that adds thinking and kind of reasoning capabilities into the Llama architecture. Box was an early access partner of this model. We will be releasing this within our AI Studio in the coming weeks or so. I think the specific reference I just caught a clip after it happened was that storage systems will become intelligent in the future. Box was sort of one of the key examples of what that will look like. It was cool to get that reference of Box in there and very consistent with what we know is happening. Great to have Jensen also have visibility into that. Ben, anything you want to talk about the partnership?
Just with NVIDIA, as they offer more models like they just announced today, of course, we're always interested in the latest, best models. This new Nemotron model is excellent. We'll be keeping an eye on that as we do with all of the other models. Maybe just to thread one needle together, NVIDIA kind of is in a series of recent announcements you'll have seen from us. Last week, we put out an enterprise eval for the new Gemma model from Google. Earlier in the week, we launched with OpenAI with their new AI agent SDK, where Box was one of the kind of initial code samples for how to build an AI agent with OpenAI. Just a week prior to that, we were one of the launch partners with Sonnet 3.7 from Claude from Anthropic.
What you'll see is maybe partly by virtue of us being an open interoperable neutral platform, we can work with all of the AI providers in a fairly neutral way. Also by virtue of us just being right at the center of Silicon Valley, we tend to be able to kind of have strong relationships at the various AI Labs. Our whole strategy works because we can be the connection point between every enterprise on the planet and the leading AI with their unstructured data. You'll increasingly see that you'll have the AI Studio from Box be this place where you can pick and choose different models from different providers and build agents around those models. I'd contrast that with if you go to each of the hyperscalers individually, their AI Studios work for their models.
They're just tuned for those particular model architectures, whereas the Box AI Studio will work with really any of these leading partners. This lets us play a kind of a neutral position, but then give customers the advantage where they don't have to move their content around every time there's a new AI breakthrough. The impact of that is one example is there's a lot of innovation in these reasoning models. Customers might one day say, "Okay, I want to use Claude 3.7 on my contracts. Okay, there's a new breakthrough from GPT 4.5, or there's a new Gemini breakthrough. I want to be able to test that on my same contracts." Think about any other architecture on the planet. You'd have to port your data over to a completely different platform the moment there's a new AI model.
With Box, you keep your content in one place, and then you link up to any of the leading AI kind of breakthroughs. You're just going to continue to see us drive this. NVIDIA's announcement today is just one more example of that.
Great. If you can raise your hand in the audience, we'll get you a mic. Please state your name and firm name.
Thanks for the question. Josh Baer, Morgan Stanley. I wanted to ask on the margin targets that they remained unchanged. The annual expansion is terrific. I wouldn't really be looking for those targets to be moved higher. The question is on your internal use of Box AI. Through the presentations, we've seen some really strong use cases internally driving a lot of efficiency. Where do we see that show up? Presumably, the last time those ranges were set out, you were not fully thinking about this level of use of Box AI.
Yeah, maybe I'll do a little bit of philosophical color, and then anybody can build on it. I think what you're going to see is that the scope of what we now can deliver has gone up as a result of the same op margin targets in that long-term model. It is almost actually kind of necessary to respond to this opportunity that we can have more output. I take, again, just the past two weeks alone, if they are representative even remotely of what the future looks like, we have been working 70-80 hour weeks for anybody involved in anything related to AI in the company. Just the output that is necessary at this point, given the sheer rate of innovation, is incredible.
We have the opportunity to now be at the forefront of that. The only way we can do that is if we max out on productivity from our workforce. AI is one of those tools that lets us do that. If you think about it in Olivia or Mark's world, the need to ramp sales reps rapidly to be able to go and sell the latest products. If you think about product design, where you can instantly prototype new products so we can accelerate our internal kind of ideation process. If you think about engineers that need to be able to code much faster. These are the things that actually have to show up for us to now make sure that we're continuing to lead.
We'd rather trade. I think you could easily be like, "Okay, could you get one or two more points of op margin from that?" or would you rather reinvest those dollars back into being in the leading position? Right now, we're making the choice to say, "Let's reinvest those dollars." Maybe we'll show up here in three or five years, and there's a different conversation. Right now, the opportunity is so big that we're going to need the productivity gains. We'd rather take those productivity gains and make sure that we're driving the strategy with that.
Yeah, and just to build on that and kind of the philosophy, I would say that you could really think about it, because it is pretty early days in terms of how this all plays out over time, as increasing what we can deliver, as Aaron talked about, and increasing our confidence in being able to deliver higher growth. For example, again, if we can have each of our average account executives spending 20% less time doing the things that they can now automate or do dramatically more efficiently, we would much rather have them selling 25% more and hiring just as many people as before and driving growth. The same thing holds for engineering. I think there's probably a point in time when that might not be true. For certain companies, that might not be true.
Again, given the choice point of, "Do we save a third on engineering expenses, or do we deliver 50% more output?" we would take the latter, especially given just how exciting this opportunity is for us in this moment.
Hi, good afternoon. Jason Ader with William Blair. Dylan, I just want to make sure I understand the monetization of AI. You have Enterprise Advanced, you talked about. You have the units, AI units, which is like the platform, right? Then is Box AI a third monetization element? Is there a way to monetize it for other parts of the suite or other parts of the portfolio? Or is that not the way to think about it?
I would say that there are AI capabilities of different levels, volumes, capabilities embedded across our plans. Just as historically, and a lot of the kind of uplift that we saw in Enterprise Plus adoption going back a year ago was driven by customers clamoring to get those AI capabilities. You can think about Box AI as extending throughout what we offer. For some of these use cases where they might show up and be monetized as AI units, which is kind of a newer convention for us, that's where customers would be able to bring in some of these capabilities. The fundamental plan would have to correlate with what they're licensed for at the core level, right? If they're an Enterprise Plus customer, you wouldn't be able to use AI units to functionality that only exists in Enterprise Advanced.
Across plans, over time, you can think about, "We certainly want to give our customers the flexibility to be able to kind of incorporate AI use cases and everything that we do into how they're using Box."
Okay, and then one quick follow-up, somewhat unrelated, but for NRR, you talk about getting from 102 to 105 to 110. Can you kind of bridge that for us? What are the components of getting from 102, let's say, best case, 110? What are those components?
Yeah, I think probably the difference between the two would be a little bit different and also maybe a function of time. You could think about the biggest impact of being around stronger pricing across our customer base and just the type of impact, not just converting into Enterprise Advanced, but a lot of the capabilities and ways that you extend in addition to that 20%-40% uplift with the newer capabilities that we've been talking about, i.e., represented through AI units today. I'd say that's the biggest driver. As mentioned, we do expect to see an improvement, not returning to the levels that we were at, call it three years ago around seat growth, but do expect that to contribute more to the business than it has this past year, for example, just not as material as the pricing side.
I think when you think about the difference between the lower end and the higher end of the range is largely a function of those same drivers, but just what is the magnitude of the impact that we can ultimately drive. I would say with the biggest opportunity and lever in that as being the all-in impact of Enterprise Advanced.
Thank you, guys. Brian Peterson from Raymond James. A lot of details on the partner ramp. I'm just curious, as we think about that 20% growth rate, how do we think about the trajectory of that? Is that something where we would think that the growth rate is higher near term and then kind of scales as it ramps up? We just love any perspective there.
Yeah, maybe Olivia or Mark, you want to take that? Dylan, if you want to provide any follow-up first, but Olivia.
Yeah, I would say at a high level, I mean, generally, and it's intuitive that when things are at smaller scale, it's easier to have higher growth rates. I would actually say this is somewhat in the same category as Enterprise Advanced and the trajectory where just due to the nature of these relationships and what you need to do to enable these partners for success, you do expect there to be a bit of a lag time, as it were, or take some time to really start to see the full impact as we onboard different customers and as they're kind of refining their practices and how this translates.
Yes, once we kind of get those partners onboarded, we would expect to have a pretty big impact out of the gate, but not for there to be a significant slowdown, just given how early we are and the type of opportunity that we're looking at. If you're thinking about what can that grow at two years from now versus five years from now, we're not going to be subject to the law of large numbers in the partner ecosystem anytime soon.
Olivia, anything to build on that?
I'll leave Dylan to actually how the model works, but just in terms of on the ground, how we're engaging with partners. It's very much the case that we are at this point going after deals that we see together that are in the market together. These are high-value solutions, right?
We're putting in the time and effort to understand how we're going to implement it. Then we want to find repeatable use cases where we can go and scale it, right, in an at-scale fashion. What you tend to see as software companies build out partner ecosystems is they find a groove, and then they get the flywheel going, right? That's what we expect to see. Obviously, we started that momentum last year. We had some deals in the market. This year, I think we're starting to see where we're really resonating with the partners, both where they can make services dollars and we can deliver on the value prop. I'm very bullish that we'll be able to really scale it in the out years.
Noah Herman with JPMorgan. I think historically, the growth contribution from new customers has typically been in the mid-single-digit range. With the NRR being 103% last quarter, the revenue guidance and constant currency being 6% for this year, that would sort of imply that that contribution would be perhaps in the low single-digit range. Is there anything, for any reason why we shouldn't assume that that shouldn't continue in the mid-single-digit range going forward? If so, what are you doing to prosecute that opportunity?
Yeah, we do expect and have seen that being in kind of the normal range. I would say on average, probably four to five points to growth. We have seen it be a little bit lower, a little bit higher. I wouldn't call it anything significantly different for this year.
Although certainly, as we think about, again, kind of where we're going to see the signs of success, especially from Enterprise Advanced and AI, we think that for over the medium term, and certainly as it relates to anything this year, we would expect the majority of that impact to be on our existing customers. As a proportion, net new might take a little bit of a step back in terms of where we prioritize. Steady state, I think we have a larger opportunity than ever to grow our contribution from new customers. It might remain kind of same proportion over time, but would expect the contribution from all of those things to be increasing in kind of raw growth terms.
Great. Steve Enders from Citi. I kind of want to get back to Jason's question from earlier. I guess as we think about the growth outlook that you're talking about here, I mean, how much of that is the macro needs to improve and we kind of need to get out of the stickiness we're seeing today to make that happen versus maybe what's more of just a timing element of some of these capabilities rolling out or some of these plans rolling out?
Yeah, I would say that actually getting into that range and our confidence in delivering that double-digit growth isn't macro-dependent. I would say that the kind of strength of the overall economy over the next several years is more determinant of maybe how quickly or at what kind of part of that range we're ultimately growing in that time frame.
Just given the opportunities that we really have a lot more direct control over than the overall IT spending environment, we feel good about that. To put another way, we're basically assuming in that that the macro does not significantly change for better or for worse. Ultimately, that could be a tailwind to growth, but something that we're very, very focused on.
Hi, Rafi from Emmett Partners. Thanks for doing this. Appreciate it. When we became investors, I guess last year in June, when I was telling people about Box, a lot of people were like, "I'll tell everybody why I think Box is a great idea.
I like to do that." Most people look at me quizzically and be like, "Box, are you sure?" I'm like, "Yeah, I'm pretty sure." I even sent him an old 2018 video of Chamath telling you guys it was 10x. He was a little early. He skipped over NVIDIA. We're going to prove him right. No, no, I agree with that. I guess I have a lot of questions, but I'm going to stick with one. The biggest pushback I got was, "Aaron doesn't seem that incentivized by the stock because he always turns down stock-based comp every year." I was happy to see in December you finally decided to change that. Why did you previously say no to it? Then why all of a sudden you say, "Yeah, actually, I want to finally make a lot more money?"
Thank you. What an interesting journey on that question. That's a very leading question. I mean, in the past and even to today, I feel well financially incentivized to make the stock work. As a co-founder of the company, not only just emotionally so tied to the company, but financially incredibly tied to the company. In discussions with the board, we kind of decided to have a little bit of a program where, based on different stock levels that are achieved, that we think make a lot more sense given the AI push, that that would unlock certain kind of grants. It felt like a comfortable time to be able to go and do that, especially given the opportunity ahead. Just in the past context, I've been very, I think, financially motivated to make the company perform well. Yeah, the opportunity ahead of us right now is massive.
Can I ask this one quick, more specific question? The AI unit thing seems so interesting to me because it seems to change so much of the way you're going to monetize your customers. And the math you gave was, I think you said it was 5% today is something equivalent to some form of consumption. Is that right? Did I see? Yeah, growing to. And you were saying growing to 30% in, I guess I don't remember what year you said to 10%, but you're just growing 30%. So the math doesn't really check out to me. It seems like almost should be much higher as a percentage of the company if it's growing at that rate.
Yeah, I think you're thinking about it the right way. I mean, that is the challenge of applying a kicker to a range of years, three to five years. We'd say at least 10%. I would also note that if you're running the math, the denominator is also growing over that period as well that you factor in. We would expect the size of our platform business to be a lot larger than double what it is today to drive even a 10% growth, much less a greater than 10% growth. Part of it was, yeah, just in terms of that there's not a specific year. Kind of running a kicker would be a little bit more challenging there.
Thank you. A question from online from an investor. It's a question on the e-Advance renewal cycle or e-Advance upgrade cycle. Does it typically have to happen at renewal? Or could you give a little bit more detail on that sales motion?
Yeah, maybe Olivia and Mark.
Yeah, I mean, we're believers in subscription software that it's all about making the customer successful. If we do a three-year contract with a customer, in our minds, we have 36 months to make that customer successful, and we have 36 months to sell them more seats and to put them on an upgrade path. We do not wait for the renewal date, whether it's a one-year or two-year or three-year deal. We are always caring for that customer, always making them successful, and always looking for opportunities for expansion. That's the way we think about it.
Lucky Schreiner with D.A. Davidson. Maybe with the growing contribution of consumption revenue to the model, how do you feel confidence-wise, just in terms of the variability of that over the long run? A follow-up to that is thinking about sales compensation, tying a little bit more to consumption revenue. What are you taking into account there? It can be a little bit of a tricky transition from a subscription-based comp.
Yeah, the one nuance, at least in terms of how our consumption revenue works today and for the foreseeable future, is it is not a pay-as-you-go, get billed in arrears for what you used model like you would have with AWS, for example, or a bunch of other ISVs. While you are buying a volume of units, whether it has traditionally been API calls, now AI units, you are just like a seat-based subscription.
Our customers are committing to a certain volume, typically, especially for enterprise customers, paying for that in advance. Then maybe they'll expand during the course of that, as Mark mentioned, tune it at the time of renewal. It is still recurring revenue. Structurally, not much different than our kind of seat-based subscription business. Although certainly, if certain use cases expand dramatically or end up kind of falling away, you'd probably see different trends and maybe variations in the net retention rate, as it were, of that part of the business. Do not expect that to have an overall material impact on our predictability and revenue visibility.
As it regards to the sales comp plan question, as Dylan explained, we do sell these in packages. It becomes a recurring revenue, and the seller gets comped on that recurring revenue. You do, of course, have some scenarios like a large migration of data, right? That's a one-time. Really, the seller is incentive to drive that migration because then it becomes basically the foundation on which they're building this value sale into the customer environment.
Great. I do have a question from Pinjalim Bora with JPMorgan online. The ability to do deep research on enterprise content is definitely very interesting. I wanted to ask you on the cost side of things. When someone engages in deep research, which I'm assuming is using a chain of thought reasoning, how does that cost compare to normal Q&A? Is that a slightly lower margin versus a normal Q&A? Will you be monetizing deep research within Enterprise Advanced or as AI units?
We're going to bring Ben back into the conversation.
Yes, to the question, when you do something like deep research or some of the other more agentic capabilities, it does mean more inference. Where a traditional call would be a relatively limited number of tokens that you go through process, when you do the deep versions of these, like deep research, you process it. The AI agent creates an outline. It goes through. It double-checks its work. That is why it takes longer to respond. In terms of the pricing model for that one, we have not yet announced exactly how that will work. You can see the different models that we have used in the past, seat-based and in AI units. We will be announcing that as that becomes available.
Maybe the only thing I would point back to in Ben's presentation is you saw the cost curve on the AI models going down. This is sort of one of the amazing benefits of AI right now, that we'll get costs that will drop at a per token inference level. We'll find more ways to still use the AI through things like deep research or whatnot. Again, that becomes another flywheel because the cost will just yet again drop. With what you saw last week with Gemma as an open-source, open-weight model, we put out an enterprise eval, as I mentioned, where it performs almost as well as Gemini in critical enterprise document use cases for kind of accuracy of data extraction. We have this incredible situation, which is we saw it with DeepSeek. We're going to see it, I'm sure, with whatever Llama comes out with next. The cost is just going to continue to drive down, gives us more choice of infrastructure.
We can use more AI in our product for things like deep research or more agentic workflows. We can basically kind of make it all work within the margin structure.
Also, the concept of AI units were future-proofed for exactly this, right? We wanted to give maximum flexibility to our customers to be flexing into these more intensive AI activities. We also wanted to make sure that they had more "standard" AI, API calls also in that same pool, right? We did not want to have them feel that they had almost stranded usage pools of AI. That was the spirit in which we created those AI units to allow them to flex into different types of AI usage.
Amarish Mehta, Tenor Capital Management Company . I know you guys, your long-term plan assumes no change in macro, better or worse. Let's say we do go into a recession in six or nine months, given what we're seeing and the markets are sort of signaling right now. How do you think that impacts you guys? Both puts and takes. Obviously, less seat opportunities, but there may be some more efficiency opportunities and takeaways from competition. Just walk us through that, how you think about that.
Yeah, I would say, I mean, even if you look at and compare how kind of the model has changed today versus two or three years ago, when it was obviously a much healthier software spending environment, you can see that the biggest part of the model that was negatively impacted was our seat growth.
Where we saw relative stability was actually in signing up new customers, a little bit impacted, and then continued strength in pricing, which you could argue that in a stronger environment would have been even higher. That has actually remained pretty consistent over the past few years. Similarly, and kind of as you mentioned, would expect that seat growth to be the part of our model that's most significantly impacted. I mean, depending on the severity and duration of the recession, you might see that start to show up in certain companies and churn if they're going out of business, for example. If you think about our full churn rate today, it is 3% annualized. A few years ago, that was 5% annualized.
In a worse economy, we've actually improved our full churn rate, our stickiness, just because our product has gotten that much stickier over time. Again, and kind of as I had mentioned, I think we're relatively resilient, just given the nature of the model and our customer base. Certainly, would see some impact and seats tend to be tied to overall hiring and things like that that would flow through into our business.
Aaron, for you, excuse me. Why do you think enterprise AI has been sort of struggling to take off? That's like the first part of the question, not at your business, but just broadly. You alluded in the opening remarks to kind of a debate on how AI impacts just the software business in general. I assume you're talking about sort of the UI layer of software. Can you expand upon that as well on that debate?
Yeah, sure. Do we have another half hour available? Maybe I'll try and do a quick, I mean, neither of those questions offer a quick version. On the first one, I think I kind of would almost sort of take a little bit of exception to the enterprise adoption side. I think what people often mean by that is things like these KPIs, like the Microsoft Copilot revenue or sort of other maybe anecdotes like that. I think that's actually just a pricing packaging problem that they had. We knew it the moment that it kind of came out that way that was not going to work. AI is something that needs to be baked into software and into technology.
It can't be seen as sort of this add-on thing that you just kind of treat this as this other kind of separate product. I think it's a little bit of a tale of two different worlds on AI adoption. I mean, if you go to any knowledge worker in the enterprise today, they're definitely using ChatGPT, or they have used ChatGPT. I think we're looking at the KPI of a couple of these add-on type businesses. I think you need to look at more just what's actually happening in the workflows more and more. That's been the first thing. We see that within Box AI.
We decided that we would bring it to all of our plans because to us, to some extent, AI is almost a reinvention of the company, which is if we were to start Box in 2025, we would not start it as a non-AI company with AI on top. It would be an AI company that lets you work with your content and your unstructured data. You would like to log into Box, and there would be AI right there. It would not be a thing that you think about as a separate kind of capability. For us, because we have some history that we have to evolve from, that is going to happen in a few parts. By the end of that journey, and you will see some major things this year that kind of represent that, Box will be an AI-first software company through and through.
That will mean everybody adopting Box is using Box AI kind of almost inherently. That is kind of one meta point. On the future of software, I think there are multiple kind of interesting cuts of this. One is, and we spend a lot of time debating this in product, and Dave or Ben, feel free to build. One is, on one hand, we see these sort of agent or chat interfaces. One thing is, maybe I will just go to those interfaces in the future and just talk with them and get my data back, which we expect will definitely happen. Then you all of a sudden have these repeated things that you do in business. Like, I want to go to the sales dashboard and see how my sales is performing.
I want to go to the contracts dashboard and see where my contracts are in a workflow. You don't have to reprompt the system every single time just to get that information. That sort of almost then, by definition, means we'll always have some version of software that you go to, you look at, you can touch and feel. Maybe there's agents that are kind of right next to that software automating parts of it. I think that this will be an evolution of what does software look like. It won't be as extreme as I think maybe some of the more hyperbolic kind of views.
We're actually fine either way because the sort of management of the underlying unstructured data that runs your business, whether it's an agent pulling that data out, whether it's an interface where you're accessing it, whether it's our agent or some other agent, you're going to need the security, the permissions, the controls, the governance, the access, the embeddings on that data. We have a lot of range of motion. This is a little bit more just like pure philosophical of what happens in 10 years. We're kind of betting that it'll be a little bit of a hybrid in the future.
I'll build on this. I think the future of software is with agents. I think it's going to be not the winner takes all. Because there are so many companies now that have access to provide intelligence, in a way, it has somewhat opened the game. The hyperscalers are not guaranteed to be the only ones winning the vast majority of the licenses and the seats. Many companies have an opportunity to provide very valuable components, for which it's essential to have a platform that is much more agnostic than what the hyperscalers offer and provides the ability to work with many. We are very well positioned for that. We have the ability to work not only with the hyperscalers, but also with any solution that appears, in some cases, out of nowhere that actually solves difficult problems. Agents, not one winner that does everything. The ability to work with a platform that connects with all, I think, are the future.
All right, great. Steve Enders from Citi again. I guess I want to ask about the AI use cases that you're talking about. It seems like a lot of them are maybe encroaching into other categories of software spend or maybe where there are some existing vendors today. I guess, how do you think about the Box ability to go capture that use case versus maybe some other vendor that's working on their own AI initiatives and kind of already owns some level of that workflow?
Yeah, first of all, the only thing we think about when we think about AI is AI and content. That already kind of limits the playing field pretty dramatically. You sort of start to say, well, how much of the workflow or agentic parts of the workflow need to be next to the content versus content is one part of 50 steps in that workflow? That kind of gives you a little bit of a continuum. Maybe someday we'll kind of publish quadrants or something that kind of shows everybody what this looks like. Just to give you a range, if you looked at a client onboarding workflow, there's probably a CRM system. There's probably an ERP system. There's some email marketing system. The documents you get from that workflow are a few things of many that you need to collect. In that case, we have partners like Salesforce, where Box AI is running on the ingested documents coming into that workflow.
We're feeding that data back into Salesforce for that customer to run the full kind of client onboarding process. We're getting paid either for the seats that the user is looking at the data on the other end or the AI units that are processing that content. In either case, our depth of knowledge about the content agentic use cases, how do you extract the data from the content, how do you work with every file type that might be coming inbound, how do you connect up to their various systems, that's something that we're really, really good at. The customer implements us inside of that Salesforce workflow as this example. On the other end of the spectrum, if a customer comes and says, listen, I have 100,000 contracts. I don't really know what's in the contracts.
It's a bunch of data like renewal information, the parties I'm working with, the amounts. I want to have a dashboard that gives me intelligence about all my contracts. I want to be able to log into it and see different trends in my business. All the data literally is coming out of the contracts. That's inherently all content-oriented. You have to be able to process the data. We need to be able to have dashboards and some workflows around that. That's something where we do need to win that. We need to be the kind of full-stack player in that example. There's kind of an even amount of workflows on both sides. We are in the no-code app business. We're in the workflow business. We're in AI to power all of those natively within Box that are very content-centric.
We have our APIs and our platform. We are working on more ways for agents to talk to each other that will power all of the use cases that sort of span multiple systems.
Great. That is it for Q&A. I will turn it back to Aaron to wrap up.
Great. Awesome. Thanks for joining us in person and attending virtually. We appreciate everybody's time and partnership with us. Feel free to reach out with any questions. Looking forward to having all of these new products come out to market and looking forward to having everybody see them. Thank you.