

Stripe’s view on how AI agents (“agentic commerce”) will rewire internet commerce and payments, plus implications for cybersecurity, business models, and company strategy.
John Collison (Stripe cofounder; 17 years at Stripe; leads thinking on payments, AI agents, and security). Molly O’Shea (host/interviewer; introduces and questions Collison). Paul and BrowserBase are referenced (BrowserBase is discussed as an example of “computer use” tooling).
Web forms are low-value; agents with stateful context (shipping/payment) will renew tasks like driver’s licenses. “Computer use” is the backward-compatibility layer for real-world systems. AI will shift commerce discovery to long-tail merchants and force aggregators to adapt. More breaches are likely as offense improves faster than defense. Stripe is preparing via acquisitions and internal AI tools.
DMV/MCP analogy; sales/legal teams building agents using internal data; usage-based billing (Metronome) and multi-model routing (OpenRouter); Stripe’s 2025 $1.9T processing figure.
Written by AI. May contain mistakes.
John discusses the evolution of Stripe and the impact of AI on payments.
“And now, there's just like some small changes happening where, I mean, leave aside everything that's going on in stablecoins and all that other interesting stuff.”
John explains how personal agents will change the way commerce operates.
“Like, the first order of change is that we think people's agents are going to be doing a huge amount of the commerce for them.”
John addresses the increasing cybersecurity challenges as AI evolves.
“We have invented a new technology that is available to defenders and available to attackers.”
John shares his optimistic view on the singularity and its implications for business.
“So you've talked about singularity a bunch.”
John discusses how businesses must adapt their strategies for the AI-driven landscape.
“And so we don't see the singularity in any kind of a pessimistic or kind of worried manner.”
John outlines Stripe's recent acquisitions and future-focused strategies in the AI space.
“And so now you're working with this new chess position.”
Learn how AI is transforming businesses and Stripe's role in it.
“And as far as we could tell, they were the furthest ahead in delivering that marketplace to help businesses actually use multiple models.”
Explore the strategic considerations of building versus buying technologies.
“But it's a little bit like that on the build versus buy where you should buy the companies before you're better off buying them than building it in-house.”
Discover how Stripe is leveraging AI internally for workflow improvements.
“And again, they're not using an application that we built for them, they are building their own application.”
Understand the potential future advancements in AI applications.
“that it can actually be the next wave of AI progress.”
Learn how professionals keep up with AI advancements and research.
“How are you staying on top of all the research, all the new advancements?”
Explore analog methods that help maintain focus in a tech-driven world.
“And so that also keeps me pretty fresh and stuff.”
Discuss the pivotal questions that need to be asked in the AI era.
“And so what is the new equivalent of bot detection for the world we are going to, where there There are the good bots, and then there are the bad bots, and you need to be able to screen between them.”
John shares his key lessons learned from founding Stripe over the years.
“but over 17 years as a founder, what have been your biggest lessons?”

Get the gist, explore an idea or ask about something said.
On Sourcery, what are the key points of this episode?
Automatic transcript. May contain errors.
0:01Please join me in welcoming to the stage Molly O'Shea and John Collison. John, thank you very much. And Paul, thank you very much. I hope you guys are all having a really great day enjoying everything browser-based. So we're going to talk about payments, agentic commerce, and everything that's going on in the computer. So Stripe processed$1.9 trillion in 2025, up 34%. Sounds right. How are you rebuilding for AI? Look, it's really interesting right now because we've been working on Stripe for a long time, right? This is our 17th year doing it. And for many years, the payments industry didn't really change that much.
0:47And now, there's just like some small changes happening where, I mean, leave aside everything that's going on in stablecoins and all that other interesting stuff. But other than that, it's all the same old. So instinct is real, muse is real? All these things are real. You saw a significant fraction of the audience using them. And, okay, I think it turns out one, filling out a web form is not a value add activity for anyone. No one chooses to fill out web forms in their spare time. And so I think the whole conceit of these agents, And obviously, again, this is where browser base and computer use comes in.
1:26But when you have the stateful context where your agent knows things about you, like the basic one is it knows your shipping address and payment details and things like that. And then when it can manipulate a computer and go off and do things in the real world, it turns out it's much nicer to be able to say, hey, please renew my driver's license for me or whatever the task is, rather than go play with web forms. And so I actually think one big part of all these personal agents starting to work is actually giving them a meaningful amount of compute so you can go do stuff. But another one is computer use actually becoming a mature enough technology because computer use is the ultimate backwards compatibility layer for the real world.
2:06Like, how long would you need to wait for, you know, the DMV to add an MCP? Like, I'm not a betting man, but I'm betting we would have to wait a few years. But it doesn't matter, because computer use solves that. And so, again, it's the ultimate backwards compatibility there. So you've mentioned that agentic commerce is a complete rewiring. What do you mean by it's complete rewiring? A few things. One is, like, it's rewiring things at the surface level, where how businesses sell is going to change pretty significantly. Like, the first order of change is that we think people's agents are going to be doing a huge amount of the commerce for them.
2:43The second change is maybe, or the second order change, is it probably changes the commerce landscape a little bit. And so, again, I don't know what people bought in their experiences, but my experience of doing AI-powered product research is it often actually surfaces niche brands or businesses you haven't heard of. I think it's quite a good story for the long tail of businesses. I think it's quite empowering that if you have a product that actually is well-reviewed and that people like on the internet, it leads to more discovery of that. And so I think the existing aggregators are going to have to answer questions because in a way, AI discovery is the new point of aggregation for commerce.
3:28And so you probably have a bit of a turning over of the top businesses that's going to happen, where if you are selling a really good product, if you're making it discoverable, then you can kind of shoot up in the rankings. I think the analogy for this is when, like, Facebook and Instagram ads targeting was so good that it led to a whole new generation of companies that were built around, you know, targeting on those platforms and they were native to it. Think of a Wish.com or a Teespring or, you know, even Timu and Sheehan, you know, to a significant degree, grew up on those platforms and did a huge amount of customer acquisition on those platforms, and they were able to more natively adopt it than the prior generation of big box retailers who, in theory, had all the same tools available to them, but they were just less native to that kind of customer acquisition technology, and it's got to flow all the way downstream to the product creation and things like that.
4:23I feel like AI e-commerce will be somewhat similar where, first order, it'll make buying things more convenient. Second order, it'll probably actually reshape the product and the merchant landscape to favor the people who can most successfully adopt it. Well, how are you thinking about cybersecurity risk as that scales? We have invented a new technology that is available to defenders and available to attackers. and people make all the trite analogies to the emergence of the motor car. The bank robbers were the first to adopt the car to conduct bank robberies and pretty quickly police forces started realizing we really need cars to keep up with these bank robbers.
5:13And so you always have this cat and mouse offense defense game that's happening. And so in our case, the new models have really advanced cyber capabilities, and it's, I mean, somewhat intimidating because you're realizing that the standard for, I mean, the good news is that Stripe has been very paranoid on cyber for many, many years, and so it's not like this is kind of totally new news for us, but we feel like the adversary landscape is getting quite a bit better quite quickly, and so it went from something that we already had a ton of attention on to something we're spending even more attention on.
5:51And it does feel like, say, over the next five years, you're going to see more breaches than you did over the past five years. Dark. This is true. Like, there's no two ways about it. Because, again, in theory, there should be some equilibrium between offense and defense. But that's this very, like, assume a spherical cow theory. And in practice, there are lots of organizations out there that are going to be too slow in how they adopt the latest capabilities. The good news for, like, the startups in the room is obviously I think startups are disproportionately in the good spot because it is easier for them to adopt the latest models with kind of the latest cyber capabilities.
6:31And they just have smaller, more contained, more homogenous tech stacks versus if you're running some, like, 50-year-old tech stack that has grown through acquisition where you have, like, a bunch of, you know, you're running on versions of Windows that aren't even supported anymore. more like that's not a great place to be in from a cyber point of view. So you've talked about singularity a bunch. You have a different perspective on it. I think it's like a little bit more optimistic than the standard San Francisco doom. So how do you, I guess, what is your perspective on singularity? How do you make this?
7:02Yeah, you know, we've been, we kind of somehow tongue in cheek at Stripe started referring to January 1st as, you know, day one of the singularity. And so kind of like the Unix epoch is January 1, 1970. We decided that the singularity epoch is January 1, 2026. That's as good a day as any to decide that we are now in the singularity. And it's not a total coincidence that we picked that date in that, if you guys remember, it was the end of last year that I feel like for many of us, it was the holy crap moment when it came to coding models and how they're getting so good. And in particular, the thing that all the AI folks talk about is this idea of recursive self-improvement where the AI uses the AI to make itself better.
7:53And again, I think sometimes it's a bit challenging because people refer to the AI as kind of this monolithic thing. But we're clearly in a world where AI capabilities are helping with the rate of improvement in AI. I mean, just look at all the math announcements that we had this week and last week. We're in math wars now. It's just not something that was on my bingo card for 2026 that we have math wars, but, you know, rival math factions. But we have. And kind of the same thing applies to the model development where the labs are using AI capabilities to go faster. So anyway, it feels like 2026 is this year where we're really feeling the effects of that recursive self-improvement.
8:40I think we're pretty optimistic at Stripe. Like, we see the early economic indicators from this, where as a result of intelligence on tap getting much cheaper, new firm creation is way up. We're seeing the highest rates of new firm creation that we've seen in recent memory on Stripe, like right now, because when you have all these tools, it's easier to start a company. It's easier to actually make it successful. I think this technology has a lot to be optimistic about. And so we don't see the singularity in any kind of a pessimistic or kind of worried manner. It's just like we are in this moment of AI takeoff right now and you should enjoy it.
9:26And it's kind of hard to deny that it's happening. Should we be scared or is that just a marketing tactic?
9:41In a business context, firstly, I think you should be paying attention. Like, the thing that we tell businesses, you know, that we're working with at Stripe is, you know, when we're talking about the cyber stuff, all the cyber stuff is real. And, again, just as we're talking about the agentic commerce stuff earlier, the entire landscape is shifting. I'll give you another example. We spent, again, the entirety of Stripe up to this point, building for developers and using APIs as a selling point. Making the developer experience easy was what we focused on, what we spent a huge amount of internal effort on at Stripe.
10:15And this would be minute details of the API documentation or how we make upgrades easy for a developer who's migrating from one version of the API to the other. overnight, you know, relatively overnight, no one actually really integrates the Stripe API by hand anymore. You know, they send their coding agent to do it. And that's a related skill, but it's actually a pretty different skill because, you know, the developer ergonomics are different from the AI ergonomics. And so many of the things that we spend time on to make things nice for developers actually don't really matter to the AIs. And what matters is a fairly different instead of things.
10:51Like, it's very easy for them to, you know, integrate the wrong version of the API or something like this. And so we are having to kind of totally change what we're good at internally at Stripe. I think that's probably true for all the industries people are working in here, where just everyone's chessboard just got a little bit of a jiggle, and some of the pieces got knocked, and, you know, the position moved around. And so now you're working with this new chess position. I think everyone needs to pay advantage of that. I think one should worry about specific concerns. and so you should specifically feel good about your cybersecurity or ensure that you can't feel good about it.
11:26I don't think people should feel this ominous, oppressive, overbearing sense of foreboding generally. I think they should feel pretty excited to be living in this fast takeoff world. So Stripe is taking off even faster than before. You have most recently made a couple of really interesting acquisitions from Metronome, Bridge, Privy, to also Open Router. So what are you assembling there? How are you redirecting for this era? So if you look at the areas where Stripe has done acquisitions, they tend to follow that pattern where Bridge and Privy were in crypto. And again, there is a certain kind of cryptonativity and way of working in the crypto world that is different to the world that we came from.
12:10And so the folks on those teams are crypto-native in a way that it would be hard to get there that quickly at Stripe, and they've injected a lot of that crypto-nativity into the Stripe culture. You mentioned Metronome, which is usage-based spelling. Because of AI, everyone has inference costs these days, and so everyone has to actually price their product differently. There are very few products these days that do not have some usage-based component or some overage. You can sign up for ChatGPT or Claude, and they will let you go a certain amount, but then they're like, sorry, you ran out of plan.
12:41You've got to buy credits. And so because of the underlying cost structure, everyone needs to have usage-based billing these days with AI products. And then you mentioned OpenRouter. We're just, well, one, I think we're obviously pretty AI-pilled like everyone else in the audience. And what we notice is that a huge fraction of, an increasing fraction and increasingly huge fraction of businesses' spend is actually going towards just AI and inference. and that could be you're doing R &D at your company or a lot of products actually have some AI inference as part of delivering the product. And you could say, oh, we're iterating towards the single model that everyone will use for everything.
13:25I think Open Router's belief is that, like to believe in Open Router, you have to believe that more than one model will be used inside a business, like the same business, and maybe that that model will change. And so people might want to, depending on the particular workload, send things to a different model. They may want to update that when new, better models come out. They may want to, when you think about it, a really high-value user you can send to a frontier model, but a lower-value user you route to a lower-value model. And so it depends on the user's lifetime value to you, all this kind of stuff.
13:58But if you believe that you actually just need to use multiple models at the same time within a business, then you need something like OpenRouter. And as far as we could tell, they were the furthest ahead in delivering that marketplace to help businesses actually use multiple models. So Stripe's goal is to increase the GDP of the Internet. How much of this is coming from AI now? Like, I won't say the exact fraction, but a lot. And part of the reason I won't say the exact, like, partly because I don't want to, but partly because it's actually pretty hard to measure these days where there's our, you know, AI business, you know, the browser basis of the world running on Stripe, clearly, you know, at the core of the AI revolution, you know, allowing people to actually, you know, allowing agents to actually use computers.
14:49That's pretty clearly in the AI category. There's a lot of businesses out there that are transforming their business with AI. They grew up in the pre-AI, you know, Meta runs on Stripe, grew up in the pre-AI era, pretty increasingly becoming an AI company. And so we find it very hard to segment companies into, you know, AI versus not AI, because if you're not using a lot of AI in your business, you're probably not paying attention. In terms of acquisitions and strategic thinking about the business and growth and new products that you're offering, when do you decide to build versus buy? that is one of the you know in the early days of Stripe I you know would go meet with the leaders of much larger and kind of further head companies and ask them questions like that and pretty quick I remember asking you know actually I remember asking Aaron Levy of Box at one point asking how product prioritization should work and he was like yeah you make a big list of all the product things you want to do and then you do the things that will be most impactful and they're like, thanks, okay, it sounds very actionable.
15:54But it's a little bit like that on the build versus buy where you should buy the companies before you're better off buying them than building it in-house. That is where I think all the taste and nuance is. I think we have tended to buy in the past when it is something that's really in the core of Stripe. We're probably more likely to be good at doing it ourselves, whereas something that is a brand new muscle, like token routing, I trust that Stripe could learn to get good at that, but it would take much longer. And then the question is, how fast is the market evolving? And I mean, open routers market is totally different today than it was 12 months ago, and it's going to be totally different again 12 months in the future.
16:42And so I think that combination of, like, a new and really fast-evolving market, oftentimes by the time you build your internal efforts, it'll be a little bit out of date or something. What have been the biggest transformations with AI internally at Stripe that you've been most excited about? You guys have all probably seen or experienced what they refer to as the jagged frontier, the uneven performance of AI where it's really good in some domains and really bad in others. And for example, if there is something where there is a huge corpus of data online, then it probably does reasonably good with it.
17:26There's many areas where it doesn't have access to any training data, and then suddenly it just kind of has to reason from first principles and it's much poorer. And then the organizational equivalent of that is like, yeah, you can do stuff in your random consumer AI app. That's not what you want to do in most business contexts. You want to actually ask it a question that has some business context. And so the reason I say this is we built our internal tool. I think a lot of companies have them, but it's been incredibly useful where you give the AI application access to the internal data, which is quite complex because you need to have all the privacy guardrails.
18:05You need to have all the access control guardrails where some people need to have access to the HR database, but everyone should not have access to the HR database. And so the permissions gets pretty complex. Just the correctness gets pretty complex. But once you have that, now you have an AI app internally that can work on the internal data sets. And that becomes pretty powerful. And, for example, all our sales teams just completely change their workflows where they can self-serve all the stuff they need to, you know, if you want to have a personalized pitch for a customer, they are building their AI applications to put together, okay, what is the data that will actually be of interest to this particular customer?
18:49And again, they're not using an application that we built for them, they are building their own application. And same with like, okay, the legal team, give you an example. People are always writing new laws, it turns out. And, you know, we operate all around the world. And so it's like, oh, this new like 600-page law in India just dropped. Have you read it yet? It's like, sorry, no, I'm reading the 600-page law that dropped in Finland. And so they have a lot of stuff to keep up with. And so someone on the legal team builds their agent for ingesting all the new regulation that comes along that might be subject to Stripe.
19:24So anyway, we're seeing a lot of people on the ground at Stripe building their own applications. And again, I think this was always the dream of technology, and people have been talking about personalized software for decades. And there's something about this moment in time where finally the barrier to entry has gotten low enough that we are now seeing a huge amount of personalized software. It's seen by the computer use. That's one of the reasons I am so excited about what companies like BrowserBase are doing is because I feel like we're at this moment in time right now where computer use is finally getting good enough that it can actually be the next wave of AI progress.
20:02Like you think we had transformers, and then using transformers we had LLens, and then with LLens we added RL, And there have been these kind of discrete hops of improvement in the development of AI over the years. And my perspective, at least, is that good computer use will come to be viewed as one of these next hops. Because it is the backwards compatibility layer on the rest of the world. Again, you don't have to wait for the DMV to add an MCP server. It's just like, it's fine. We're just going to computer use it. And so I feel like that is going to be the next wave of AI applications is taking advantage of this new capability.
20:39How are you staying on top of all the research, all the new advancements? Like, how do you keep your information diet rich? Same way everyone does. I know all these people who, you know, these people are like, oh, I read like 10 AI papers before I go to bed. So I'm not doing that. But I'm not sure they are either. But I vibe code my applications that I want to use, and that's very enjoyable. And again, you can actually get significant utility out of stuff these days. And then just constantly talk to Stripe customers about what they're doing. Again, it's helpful that we try to pay a lot of attention to the companies that are fast-growing on Stripe and make sure we're spending time with them.
21:26because, you know, someone like a, you know, browser base is well on its way to becoming a huge Stripe company or lots of other kind of early stage startups like that. So we've always tried to, you know, look at where the puck is going rather than, you know, company's current scale. And if you are spending time with the fast growing companies on the platform, they will invariably be in AI or kind of around it in some way because that is just the fast growing area. And so that also keeps me pretty fresh and stuff. Okay, well, maybe this is a more abstract question, but is there anything that's more analog that has helped you kind of stay focused and not get AI psychosis?
22:08Who said I don't get AI psychosis? What is more analog that helps me stay focused? A favorite book, movies, normal human things. All the normal human things, like getting out in nature, like putting down the phone from time to time. I don't know if people have tried that, but it's pretty nice out there in the wider world. Like, you know, we're here in California. It's like one of the nicest places in the world. And so you just like go across the Golden Gate Bridge to Marin or something, and that's pretty restorative pretty quickly. If people are visiting from out of town, you got to get like outside of SOMA.
22:48So if you're like coming to this conference and you're staying in Union Square or whatever, The key thing is you've got to get outside of this one block radius, especially while Dreamforce is on. I mean, good God. And so that is my main encouragement to it. It also will help with avoiding the AI psychosis. I mean, it does surprise me, because I visit here quite often. I don't live here, that people come here as tourists. But it's beautiful. I know. Of course I do. I know. But, like, I'm just used to coming here, and it's like an office park. Yeah, yeah, yeah. But people come here because of culture.
23:22Yeah. I don't know if I'd say culture, but... History. Yeah, exactly. Nature, yeah. So what do you think, maybe this is a little bit more of a tricky question, but what do you think are the main or one of the questions that isn't being asked right now in this era, AI, agents, computer, that we should be asking? One, like I said, is... I think this is being asked, but I think it's going to become a much louder and more pointed question, is, for example, many sites do bot detection and bot prevention, and that is a reasonable thing to do, or it's understandable how they got there, but it's increasingly all bots.
24:06And so what is the new equivalent of bot detection for the world we are going to, where there There are the good bots, and then there are the bad bots, and you need to be able to screen between them. And so, again, I think we are just at this really interesting moment happening now where, again, as I see it, you really should play with one of Instinct or GrokBot or Muse from Meta or any of these kind of AI assistant tools because they're a very powerful sense of what is coming and that new modality for how AI products will work. But again, I think they ask a lot of questions about what the new equilibrium will be, and I think we just don't know that.
24:52As we close out, I have one last question, but over 17 years as a founder, what have been your biggest lessons? Look, I think if you're really looking to condense the Stripe story or Stripe experience, One, I'd say there's no shortcuts.
25:18Stripe never had a South by Southwest breakout moment. There were times when it would be flicking the growth dial and saying, is this thing on? And just it was growing. Every year has been a bigger year than the prior year. It turns out you just do that enough years one after another and you eventually get to$1.9 trillion. But it never felt like, oh my God, this thing is really working. It just felt like, yeah, I mean, growth is down a bit from last year, but in absolute terms, it's bigger. And you eventually get there. And there's a lot of years of that. And similarly, we'd invest everything.
25:52I mean, this is obviously kind of definitially true. Every new product you start is kind of immaterial and a toy to begin with. But we rarely started products that felt like the future of the company at the time. We just started a neat thing that we thought delivered value for customers. and so I think there's a real element of Stripe that is just stick with it-ness where we didn't quit and we didn't get acquired and we just stayed plugging away at it and I think that's underrated in a lot of tech stories. NVIDIA is the largest company in the world and they've just been making GPUs since the 1990s and they made better GPUs and again I think they thought of improvements like oh we should have CUDA and they thought of these improvements to have but it's a real story of stick with it-ness because there were many periods when NVIDIA is much less hot than it is today.
26:39I think we always tried to play the long-term game, you know, infinite game with people where you want to do right by your partners and you want to be seen as good to work with. I think reputation matters a lot. You know, we care about Stripe having a reputation for being a good company to work with. And so I think just treating being a good counterparty seriously or something is something that really matters. And then the last one is, we talk about this at Stripe, you know, we have a set of operating principles in terms of how we work, lots of businesses have them. Our first one is just users first.
27:19And we always try to start with, what does the customer actually want? And work backwards from that, especially when it comes to product development. And I think it's very easy for companies to go adrift and start getting into, you know, you know what market we should get into? or a great new revenue pool would be X or really get into this internal company-oriented way of thinking, that's fine to do and people can like their spreadsheets or anything like that. Well, John Carlson, thank you so much. And thank you to Paul and everyone at BrowserBase and all of you for coming today.
John Collison, Co-Founder of Stripe, on how AI agents are changing commerce, payments and the internet itself.
I sat down with John at Browserbase Navigate to talk agentic commerce, computer use, cybersecurity and why Stripe considers 2026 the beginning of the “singularity epoch.” We get into why computer use could be the next major leap in AI, how agents will reshape product discovery and transactions, and why Stripe is rebuilding parts of its infrastructure for AI instead of humans.
We also cover Stripe’s acquisitions of Bridge, Privy, Metronome and OpenRouter, usage-based billing, multi-model routing and how AI is already transforming workflows inside Stripe.
Browserbase builds infrastructure for AI agents that use the web, providing cloud browsers and tooling that let agents navigate websites and take actions across existing web interfaces.
𝐓𝐈𝐌𝐄𝐒𝐓𝐀𝐌𝐏𝐒
(00:00) Stripe’s $1.9T & Rebuilding for AI
(00:31) Why Agents Will Replace Web Forms
(02:20) How AI Rewires Commerce & Product Discovery
(04:40) Why AI Could Mean More Cyberattacks
(06:49) Why Stripe Says the Singularity Started in 2026
(09:29) Stripe Is Rebuilding Its APIs for AI Agents
(11:38) Why Stripe Bought Bridge, Privy, Metronome & OpenRouter
(15:12) When Stripe Builds vs. Buys
(16:55) How AI Is Changing Stripe Internally
(20:39) How John Keeps Up With AI
(23:28) The Coming Good Bot vs. Bad Bot Problem
(24:53) 17 Years of Lessons Building Stripe