All right, that is just a preview of many of the exciting things that we are delivering at Box to help you transform with the full power of AI agents and enterprise content securely. Now, we can't do this alone. So, we have brought together a partner ecosystem to make sure that we can help transform organizations anywhere on their journey with the full power of AI. And one of our key partners is Amazon Web Services. And I'm incredibly excited to introduce Rahul, the VP of data and AI GTM at AWS to share a bit more about how we are working together. You guys have some breaking news with OpenAI and this kind of rocked the industry because now for as far as I can tell the first time ever, customers can choose a different cloud to be able to use the AI models on through Bedrock. So would love for you to maybe share a little bit about that partnership. We earlier today had Dom Gillo from OpenAI on stage talking about how we're working with OpenAI. So now there's a clear kind of joint story as well there, but would love for you to share a little bit about what you're doing with OpenAI.
Yeah, thanks Aaron and it's great to be here. Appreciate the opportunity and we are super excited about what we're doing with OpenAI. As you mentioned, it's a recently announced partnership and we're delighted to be able to bring OpenAI models and so you can get the GPT models from AWS through Bedrock. We're also working with them on managed agents and so this allows you to easily build stateful agentic applications that are powered by OpenAI on Bedrock and then OpenAI frontier will also be available on Bedrock as that becomes more broadly available. So, you know, as we work backwards from what customers need, obviously that choice of provider available on Bedrock alongside Anthropic is a big deal, as well as all of the other model choices that we offer our customers and we're excited to partner with OpenAI and get this in front of as many customers as possible. I think it'll be a big deal.
Yeah. Awesome. So, for anybody who somehow has been maybe off the grid for like the past month, very big deal. So I can as a developer or an IT organization right now or imminently I'll be able to say okay I want GPT 5.5 running on Bedrock and then go against my cloud consumption on that front.
Yes. Exactly. So you'll be able to select that as a model. The beauty of Bedrock is you can maintain stable APIs but pick your choice of model and this will now be available to developers coming very soon. We announced it just a couple of weeks ago and it'll be available shortly.
Awesome. Great. Well yeah congrats on that. And so now with our joint partnership as well, we're going to make it as easy as possible to be able to leverage Bedrock services whether it's OpenAI, Anthropic or other models. And we're doing a lot of great work with you guys on that front which certainly brings us to AI in the enterprise. You guys have a very unique kind of view into the landscape just given the scale of the workloads you see, the kinds of customers you serve kind of across every single industry. Would love a little bit of maybe state of the union where are we with AI at the moment. You know there's a lot of talk of we're beyond the chat phase of AI moving more into the agentic phase. What does that look like from your vantage point and with working with customers?
So what I see across close to a thousand customer conversations at this stage is that we're definitely in the year of agentic AI. I'd say in 2026 we're seeing active production deployments for complex end-to-end business workflows and I think that is going to continue to accelerate. And you know, one of the key things that we see in the enterprise is that, you know, the key differentiator between customers that make it into production versus those that stay at the POC phase is the integration of enterprise data with AI. And that's really the only key differentiator that customers have. Anyone can access state-of-the-art AI models. It's the combination of enterprise data about the business, about customers, all of that embedded knowledge coupled with state-of-the-art AI that allows customers to build unique differentiated experiences. And we see folks saving tens of thousands of hours a year in clinical research from things like know your customer processes in banking and finance to claims auto approvals, and insurance. It's really widely spread and accelerating at this point.
You know, I think we've seen so much adoption and so much ascendancy on AI for coding and that kind of is what we have a lot of the best proof points for. As you think about we're moving from coding agents to now more areas of knowledge work and workflows. Where are you seeing the hit rate be highest? What do you kind of see as the areas that customers should be thinking about and focused on on that front?
Absolutely. So coding you're absolutely right that's taken off it's growing exponentially still. I think we've seen a lot of early adoption in things like intelligent document processing in areas like support. But I would say at this stage it's expanding to really every business workflow you can think of and so you know as we think about it from the AWS perspective, you can really accelerate any business process with agent assistance and so we've seen things as diverse as individuals prepping for a bi-weekly review with their CEO using agents to help them prep that material. Today I'm using tools that allow me to generate a custom customer-specific presentation that incorporates all the latest information from CRM, all the latest news, all the latest releases. And so I think we're really just limited by imagination, access to data and the APIs at this point.
When you think about the areas that customers should be thinking most about where to get the most value from agents, you know I think there's always this tension which is do I use AI to take a current process and maybe streamline it make it more efficient possibly reduce costs, do I use AI to accelerate something where more output in that area could generate more value for my organization. Do you have any kind of sense of where customers or enterprises should be leaning where you've seen the biggest proof points for agentic success?
The way I think about it, Aaron, it's really about working backwards from key business objectives. So, I would say more about what is your business goal? Use that to then define what's the minimal set of data assets I need to light up that particular goal. And then the third piece is this idea that discipline equals freedom. If you set up good guard rails from a security, responsible AI and financial perspective, you can actually let the organization free to innovate. And so once you kind of have those three things set up, then it's about sprinting and really getting after it. You know, if you are early in your AI journey as an enterprise coding, agentic assistance for employees, customer support, and customer relationship management, these are great places to start. I also think AI to eliminate drudgery is a great way to get started. So, you know, we used it to upgrade our Java SDKs. Again, code that no one likes touching. AI was great at that. But if you're further along, I would look for key business objectives that you're trying to drive. Maybe you're trying to drive down the cost of originating a loan. Maybe you wanted to find a new line of business or up throughput and then work back from there.
I got out of a meeting very recently where we finally are updating a feature in Box that I don't think a single engineer has touched in 15 years. And it's not because it's sort of an unimportant feature. It's just like the tribal knowledge eventually evaporated and nobody really knew how to maintain it and so it's been on just life support and finally AI is now the reason we can just keep innovating in these areas. And so whether it's the migration project you never would have done before, it's the upgrade of an SDK, it's the new features that you couldn't have gone and triaged, I think it's this amazing relief valve in our road map and in our normally constrained processes which is incredibly exciting. And so the thing that we're excited about and I think you guys have been on the forefront on and I've heard Matt Garman and others talk about this is you know start to think about AI is really how we have this new form of abundant intelligence. How do you use it to do all the things that you never could have done before as opposed to just taking today's processes and replicating them. So you guys are right at the forefront of that and it's been awesome to see. Now there's sort of no free lunch in this world right now. And I don't even mean the cost of the tokens, but the work it takes to make AI useful and be able to get your organization into an AI ready way and format. How do you kind of see the upgrading of infrastructure, data platforms, how companies need to think about their information, their systems, so that way we're ready for agents?
It's a great great point and great call out. I do think customers that really want to embrace AI are really thinking about how to modernize their data platforms and get themselves in order. You know, and often that means moving out of legacy systems and thinking about moving into more scalable architectures in the cloud. You know, often enterprises go, 'Wait, I can't do anything. I've got to get my data house in order before I can get started on AI.' And I think that's actually a mistake. I think you have to innovate while you modernize. And so you can do things like set up MCP access to data in its legacy form, start to use that to build your learning on AI even as you modernize simultaneously and then you can just repoint under the covers. So I think one of the things that I spend a lot of time talking to customers about is yes, you do need to get your data sorted, but you don't wait for clean data. A, there's no such thing, and then B, today's technology with agents and models allows us to make sense of messy data in ways that we couldn't before. So figure out how to do both together.
We're seeing a lot of customers, I'd love your take on this, where they're ready to deploy agents. There's huge use cases right away out of the gate that will generate them more revenue and drive productivity up. But data is still stuck in some of these legacy systems on premises environments. They're not in the cloud. You know, maybe they're in platforms that are in the cloud, but they're not really built for agents. And are you seeing an acceleration to getting data ready in that kind of way? Because that's obviously an area that we're working together on that we're pretty excited by.
