Rethinking the future of digital transactions

[Music] So to round us out, uh, G opened the day by saying our goal of ship AI was to really have an applied AI conference. So everybody got here would get down to brass tax and leave even building an agent in real time or certainly knowing what you might want to leave the room and do. So, I wanted to finish today by getting some real operators or at the forefront of agent building on stage with me uh to talk about what they're seeing out in the wild. So, as I thought about who's the right set of people to bring to stage, uh what are what are the

core things you need to build an agent? Well, you need logic and reasoning, you need memory and state, and you need an ability to create and capture value. So, with me, I've got Kate from Anthropic, I've got Arjun from Neon Databicks, and Jeff from Stripe. So to kick us off, I thought it would be most interesting to get a sense from each of you. What is the most transformative agentic application that you're seeing an enterprise build at this point? There's so many. I think well I think

every presenter up here said agent at least 10 times. I'm sure every single one of you is building an agent. Um some of my most favorite ones that I've seen Rakuten has built an agent that can work for seven hours on its own. Um, so they basically set the agent. It's a coding agent. They set the agent, everybody goes home, goes to bed, comes back, and a whole bunch of work has been done. It's kind of like having a buddy on the other side of the world that you're co-working with.

There are so many others. Some of the big banks that we work with are completely transforming things like KYC. Um, big insurance companies completely transforming claims processing. pharmaceutical companies thinking about taking processes that used to take three months now take 15 minutes with agents to help them with things like regulatory document processing. It's really amazing to see how quickly this technology is changing everyone. Arjun, what what's been your top use

case? So, interestingly for us from the data layer, we've actually been very impressed with text to app agents that generate apps. It's been quite impressive. uh enterprises build phenomenal agents with uh ML scientists and devs, but agents like V 0 have impressed us because typically when you think about a database, you have a dev building a database. You put a little bit of data in test, uh figure it out, put the schema, put into prod, and it scales out. But then you have V 0. Vzero comes in and wants to build one app per

end user. And now we're at millions of apps. And each of these apps needs a a database. and they have some very very interesting use cases and pushing on the data infrastructure of the way we do it. Um, and I would say people today think of Vzero and its competitors as a little bit like a consumer or a proumer tie. I think that's shortsighted. I think Vzero is almost ready for the enterprise. The problem is the enterprise is not yet ready for Vzero, right? So, we got to put all the scaffolding around it to make it uh productive. So, I would say

we're impressed because we feel like we're at the cusp of enterprise wipe coding, if you will. uh a agree actually on on Vzero and we did not plant him to bring that one up. Um actually I I think that we're seeing some of the most compelling usage in enterprise right now. Um so that's where I've been spending the bulk of my my time. Jeeoff, what about you? I'll give a consumer one and a business one. Uh on the consumer side, many people use ChatGBT as well as other types of of consumer agents. And I've just been really blown away about how

quickly consumer behavior in terms of discovery has changed and what used to be on page, you know, 10,0001 on a Google result. It's a little funny. You you know, Google says, "Oh, hi. Uh here's a in 01 seconds I found 8,000 pages for you. Um you know, go ahead and enjoy." Whereas whereas a a prompt with with many of these LLM products can bring something that's really personalized to you immediately and now you can buy immediately inside of uh chat GBT and and f future agents um when and Stripe and and OpenAI worked to