The Agents Economy Backbone - with Emily Glassberg Sands, Head of Data & AI at Stripe

Hey, everyone. Welcome to the "Leading Space" podcast. This is Alessio, founder of Kernel Labs, and I'm joined by Swyx, editor of "Leading Space."

Hello, hello. We're back in the studio with Emily Sands from Stripe. Welcome.

Thank you.

So Emily, you're head of data and AI at Stripe. That's a, that's a big title. Uh, what does that actually mean in practice?

So Stripe is building financial infrastructure for the internet. We started out as payments infrastructure, and now we are helping businesses solve a whole range of problems. How do they, um, accept recurring payments like subscriptions? How do they do usage billing, revenue recognition, tax, money movements, accept stable coins and more?

Um, and when you think about what we're looking at, we're looking at on the order of one point three percent of global GDP, uh, about one point four trillion dollars a year is processed on Stripe. And so that obviously creates a very unique opportunity to use that data to understand both what's happening in the economy, what do our users need, but also feed it back into the product, um, to power better payments experiences, so cut down on fraud, drive the right authorization, better customer-facing experiences, optimizing the checkout suite, um, and more.

So anyway, our, our data and AI org is just really focused on helping Stripe, uh, make effective use of our data, and that starts sort of all the way at the foundation layers, right? Like what's the data platform?

How do we do data engineering? What are our, um, ML infrastructure, AI infrastructure, and then all the way up to the, to the applied layer. Um, we also have a, a fun little group. It's actually quite small. It's just two dozen people, but, um, we call it the experimental projects team.

Mm.

Um, and it's not data specific, but the premise is experimentation can and should and does happen everywhere. But there are often these sort of cross Stripe opportunities that are being pulled out of us by our users just given the pace at which the world is changing that aren't natural or easy to jump on within any one product vertical today.

And so these are just quite senior, quite seasoned engineers who run at those opportunities and, and zero to one them and get them off the ground. So our agentic commerce work came out of that. Um, token billing, which we can talk about, uh, in a bit, also came out of that and, and that's just a fun sort of side angle, um, within our group that's, that's proven very high leverage.

Yeah. I, I like the frames, framing that, you know, Stripe's mission is build financial infrastructure for the internet, and y- your subset of that is build fi- economic infrastructure for AI.

Yeah.

And that's, that's a, that's a pretty ambitious goal.

Um, you've been at Stripe four years. At what point did AI become a title level thing? Because I mean, you were obviously using machine learning for like fraud detection and everything, like you were-

Cassandra

... tell us about... Yeah.

Yeah.

Yeah. We started investing in AI or LLM specific experiences basically when GPT-3.5 hit the scene. We were like, "Okay, we need everybody to be able to have, um, high quality, safe, easy access to LLMs, not just for their own, you know, day-to-day work usage, but actually to build, you know, production grade experiences."

So that, that, that sort of... What was that? Early twen- like January twenty twenty-three or late twenty twenty-two, we, we started reasoning about, okay, like, you know, it's not just ML infrastructure, it's also AI infrastructure. It's not just ML applications, it's also AI applications.

But then it was really only in the last year and a half or so that we said, "Hey..." I mean, we had like transformers or whatever before, but only in the last year and a half or so that we were like, "Hey, we actually need to have our own domain specific foundation model, and actually we can move from these, you know, single task point solution ML models to, you know, a, a much richer, denser payments embeddings that can then power the, the various downstream applications."

So, so I think it was an evolution for us. Um-

Yeah

... and you know, I think we could still debate like what's ML and what's AI in the industry at large, but you're right that we're more than a decade into using ML at Stripe, you know, way back in the early days for not just Radar, which I think people know about, right?

Like, um, the, the machine learning systems that block fraud for our customers, but also ML internally for our own operations. Like every other payment service provider is onboarding max a dozen users a day. We're onboarding thousands of users a day, and you have to make sure they are supportable and not fraudulent and are credit worthy because you are processing their transactions, and that alone requires machine learning and, and Longhas.

Yeah. I would say, uh, you know, it's, it's kind of interesting how the in domain specific models first come out like pre-foundation models, and then we have this foundation model era. But then at the scale of Stripe, I imagine that you also have to just serve in so much volume of inference that then you might have to domain specialize them again.

There's this kind of like bell curve.

Yeah, there's fine tunes on top.

Yeah.

There's fine tunes on top. I mean, we are, we see like fifty thousand transactions a minute. And so yes-

But not all of that goes through your foundation model.

Ev- yes. Every single-

Oh

... transaction. So for example, the foundation model, one of the things it powers is detecting card testing attacks. Do you guys know what card testing is?

Yeah.

Yeah. Okay. So, um, for listeners it could be like a card tester could be enumerating through cards or they could be random guessing cards. They find a card that works and then sometimes they use it for fraud, more often they sell it.

Lots of traditional machine learning models can do a pretty good job detecting card testing, but card testers have gotten clever and one of the things they do is that they hide their card testing in the volumes of very large businesses.

So if you think about a very large e-commerce company, you can think about how many transactions are there. A card tester might sprinkle like one hundred, two hundred, three, four, five cent transactions in testing. Traditional ML is like not gonna catch that.

Then you have a foundation model, each, you know, charge becomes this like dense embedding. You start to see these clusters sort of pop out and you know in real time that they're card testing and, and you can block them.

So, uh, yes, it is, it is happening on the charge path. In less than 100 milliseconds of latency.

Yeah.

Have the foundation model enabled more data to be put in the embedding? I, I think, like, you know, things like, you know, number patterns and, like, ZIP code versus, like, location, I think those you could do before. What- are, are there any new data points that you get?

Yeah. So, so I think there's, I think there's two big things. Like, one, you know, when you're, when you're building a small model, you're usually like, oh, looking at the data that, like, has reasonable labels. It's, like, recent history.

You probably have some, like, hand-engineered features in many cases. If you're actually, like, building an FM and you're imagining there's many downstream use cases, you're putting, you know, tens of billions of transactions in it, um, you're putting, like, the entirety, like every detail of the payment in it, and, and letting the FM reason about what are the components that matter versus not.

So it is literally all the things. But I think what's even more interesting is, like, the last K. Like, like, what matters is the sequence. What mat- Like, you can think of a payment sort of like a word, and so you can think of payments data kind of like language data.

And what matters isn't the word, right? It is the word in relation to the words around it. But what's tricky about payments is you don't see, like, you know, Emily on a podcast saying 20 words and know those are the words.

Like, the sequence that could matter could be, you know, this particular retailer charges from this IP. It could be anyone on a Friday night with this credit card. And so you kind of have to choose a broad swath of relevant sequences to capture kind of the last-

Mm

... the last K that matters. So if you think about, like, like a movie, right? Like, what, what are the s- what are the scenes in a movie that you need to be watching, um, to know if there's something anomalous happening?

Yeah. And, uh, for listeners, I think you've talked about this in, in a number of places. The card-testing detection numbers went from 59 to 97%.

Yeah, on large users.

Uh, that's-

Exactly

... which sounds pretty helpful.

Yeah. It was really helpful. And but the other-

If, if you-

... thing that was really helpful was, like, the speed at which we got it out, right? So, like, we had a, we had a, um, a couple of the large AI companies came to us and they said, "Hey..."

This was, like, after card testing. They were like, "Hey, um, Radar is amazing for finding fraudulent disputes." And that's what it's, that's what it's trained on, right? Transactions that result in fraudulent disputes. But we have all these sus, like suspicious transactions, that don't result in fraudulent disputes, but we still want them flagged.

We want them flagged because even though they don't result in a fraudulent dispute, even though we get paid for them, like, they're bots, it's not good traffic, like, they're messing up our numbers, all sorts of different reasons. We can talk about some of the fraud that AI companies are facing.

And so we want you to send us a pipeline with all the sus transactions, even if they're gonna be revenue generative, 'cause we're probably gonna wanna block them. And it was, like, literally days. We're like, "Okay," like FM embeddings, clusters, you know, good textual alignments.

You can start to label them, and you're like, "This is the clusters, uh, you know, that look sketchy 'cause they're enumerating some component of the login flow. These are the, th- these are the ones that look sketchy 'cause they're enumerating some components of the login flow.

These are the ones that look sketchy for XYZ other reasons." And then the AI companies could literally say, "Okay, this batch we want to block. This batch-"

Mm

... "uh, we don't." So it just allows you to move faster on identifying, um, not just new fraud vectors, but, like, whole new types of suspicious transactions.

How has the scale changed with AI? So before, uh, you know, I used to run some software website, and we would have the same issue. People buy the software and then gets charged back, but it's like 20 bucks.

Yeah.

Like today you could like, you know, use the credit card and sign up for the OpenAI API and spend $10,000, $15,000. Like-

Much bigger

... what's the shape of the fraud-

Okay

... today?

So friendly fraud is, like, not stolen card credentials, but something like non-payment abuse, free trial abuse, refund abuse. So it's me, they're my credentials, but I'm not actually creating accretive revenue for the business. This has happened for a while, and actually if you, if you ask business leaders, like, I think something like 47% payments leaders, like 47% of them will say that their biggest fraud challenge is friendly fraud.

I would say this was, like, just much less of an issue for SaaS for two reasons. One, like, what were you stealing? You weren't stealing-

Right

... computer inference or whatever. And two, more importantly, even if you were stealing some service, like the marginal cost of providing that good or service for Salesforce or whomever was like near zero.

Mm-hmm.

And so it didn't totally crush your unit economics. Now we're in the world where GPUs are expensive, inference costs are high, and free trial abuse or refund abuse or general non-payment abuse, right? You, you rack up these charges and you never pay, is like existentially threatening for AI businesses.

I was talking to a, a small AI founder the other day, because we're, we're building sort of a suite of radar extensions that are explicitly targeted at this type of fraud. And everyone tells me it's a huge issue, and so with every company I talk to, I try to dig in on, for you, what exactly is the issue?

And this is the first guy who told me it's not an issue. And I was like, "Oh, fascinating. What are you doing?" He's like, "Well, I completely shut down free trials, and I dramatically throttle credits- ... until you've proven ability to pay."

And I was like, "Well, you don't think fraud is an issue?"

Yeah.

But it's like, totally like you're choking your own revenue, right? So anyway, we, we worked with him. We got free trials back on and, and, uh, that's, uh, that's in flight. What's interesting, this is definitely a problem for AI companies because of the marginal cost, but it's not only a problem for AI companies.

If you think about like advertisers, right? If you're like, you're like a social media platform, right? Advertisers come in, you let them start advertising. They do post-hoc billing, right?

Mm-hmm.

So you rack up some spend, and then you pay. Um, and if you don't pay, that actually is expensive, not because you've stolen compute in this case, but because you've taken ad slots from businesses who would've paid. Fascinatingly, I don't know if I should mention, I think I can say it.

Okay, so Ro- my friend got a Robinhood credit card the other day.

Mm-hmm.

Yeah? Um, and literally as part of getting the Robinhood credit card, he was marketed that he could also get free trial cards. And I was like, "Oh, tell me more. What are free trial cards?" Free trial cards are- Basically, like cards with your name on them that are good for 24 hours and then expire, so that you can sign up for free trials without ever having to get charged.

So in the hands of a well-meaning consumer, that sounds fine. In the hands of a fraudster, that's like extremely disruptive to the AI economy. We just announced our free trial offering. We can catch the majority of free trial abuse at the source.

We're working on the analog for refund abuse. Um, one of the places where refund abuse is really painful, you mentioned this in the context of large volumes, like some of these AI companies will have, like enterprise-grade plans where it's like $600 or $1,000 or $10,000 a month for hundreds of thousands of credits in whatever units they're providing, and those are the ones that are getting hit with refund abuse.

So a lot of the free trial abuse is like the consumers, the little dollars, but it adds up, but a lot of the refund abuse is like very, very large subscriptions.

Yeah.

You use it in full, um, and then you go and cancel. And we see the usage happening, so like... And we can verify that it's the person. So there's, there's a lot we can do here, and, um, I think, I think we can burn it down, but it's like clearly creating a lot of pain for the ecosystem today, and just that vignette of that founder who told me he'd solved fraud,

Wow

You know, I think that just speaks to like it's so painful, they will literally-

Right

... give up revenue to, to not have to deal with it.

I think you teased a little bit about how you're extending Radar to serve, uh, new AI business models. I, I think Stripe in general I think is interested in like enabling payments for, for, uh, these AI business models.

But basically, like what, what do people want?

Yes.

And maybe what's realistic versus what is not realistic. I don't know if that's a, a term.

For sure. So, you know, I think of Stripe as like the skeletal system for AI companies. So if you look at the Forbes AI 50, all of the Forbes AI 50 who monetize online monetize through Stripe. And what do they use us for?

You know, most of these companies, I mean, you've seen it with like the, the Cursor and Lovable examples, right? They build these very scaled businesses with very lean teams. And so they wanna go like all in, kind of on Stripe to get many layers of the sort of economic infrastructure, financial infrastructure stack in one go without needing to hire humans to do it.

So they use us for payments. Um, these AI companies are, are going global from day one. Like we were looking at the top 100 grossing AI companies on Stripe, and like the median was in 55 countries at the end of their first year- ...

and over 100 countries at the end of their second year, which is like twice as global as the SaaS wave from three years before. So they, they almost all adopt our optimized checkout suite, which comes with 100 payment methods out of the box, very global reach.

Um, they almost all adopt Radar and our fraud suite. And then one of the things that's been really interesting is, you know, I think the, the market's still trying to figure out what's the intersection between supply and demand, and so there's a lot of iteration across monetization models.

Like is it a fixed fee subscription? Is it pay-as-you-go usage? Is it this credit burn down model? And there are revenue implications. There are also fraud implications. But equally important, there's like unit economic implications, and I think one of our recent ahas was, you know, as the LLMs have gotten better, more and more AI companies are wrappers.

And I don't, I don't say that in a... Sorry, wrappers with a W-

Derogatory. Yeah, yeah

... R, not a R. And I don't say it in a derogatory way. I say it in the same way Arvind Srinivas once said to me, like, "I am proud to have started as a wrapper."

Yep. Complexity is a wrapper.

"It's allowed me to find product market fit and build an amazing product and move really quickly and not get slowed down in the research. And other people could provide the underlying models. I could do that later if I wanted or not."

But because a lot of these AI businesses are wrappers, right, their services have an inherent LLM cost underlying them. We know that LLM model providers are ebbing and flowing. The underlying models are getting better or worse. The price of those models are getting better or worse over time.

And so that actually leads to a lot of complexity in how you price your final service when that final service is like so dependent on an upstream LLM. So one of the things we launched two weeks ago now, uh, we call it token billing, but it's basically you...

It's an API that lets you track and price to inference costs in real time. And what does that do? It's like, well, if, if your service is built on an underlying LLM and the model cost drops 80%, which we've all seen it happen, right?

You don't wanna keep your price where it is because the competition's gonna swoop in. But then conversely, and I think more threatening, if the, if the cost of the underlying LLM 3xes, which it also sometimes does, surprisingly, you could have unit economics that are literally underwater if you don't, if you don't adjust your price.

So, um, you know, token billing is an example. But, but we're seeing these AI companies, you know, iterate across usage-based billing, outcome-based billing-

Yeah

... is kind of an interesting one. Um-

Does, does Stripe get involved there? Because that's not really within your normal wheelhouse.

We do. So our... We have... So we have payments, and then we have a billing suite, and almost all the AI companies use our billing suite. Billing includes things like fixed fee subscriptions, but it also supports usage-based billing, so like metered billing.

Yeah.

Um, and you can define usage in all sorts of different units, and we have a number of customers, including Intercom, who actually define it in terms of the outcome. So in this case, it's like support cases resolved.

Resolutions.

And... Exactly. And you know what? I think it's really interesting. I'm an economist by training. I told you earlier in the elevator that I think physicists make the best, uh, scientists or MLEs, but I didn't know that at the time, so I'm an economist.

Um, and I think a lot about, okay, like what makes the market efficient or inefficient? Um- And one of the things that I worry about in AI is it's incredibly hard to take a product to market when someone has to pay for it before they see the value.

Um, and that's especially true with AI because a lot of the buyers, especially enterprise buyers, don't understand how to evaluate the underlying technology. And so if you can get your foot in the door by saying not just, "Oh, you'll only pay for what you use..."

I mean, pay for what you use is kinda helpful 'cause they're not committing upfront to some huge contract, but they can come in with a fear like, "Well, what if my employees use it a lot and it's not actually helping the business?"

Mm-hmm.

And if you can come in with, like, an actual cost sort of pricing function that is clearly profit positive for them, um, it's a lot easier to get your foot in the door, right? So you know that a human resolving a support ticket is X, I promise that I will charge you, you know, less than X.

Like, it seems conditional on quality strictly better, uh, for you to try out my service than not. So anyway, we see, we see a lot of outcome-based billing. The other thing we see an interesting amount of from AI companies is stablecoins.

Hmm.

And this one's, this one's earlier, but their use cases are interesting. So, like, if you, if you go to vZero now and sign up for an account, you can actually pay, uh, in stablecoins. We're seeing this a lot for AI companies that wanna have very global reach, but also for AI companies that have very high price points.

So, like, Shadeform, the, the YC startup, is a great example. Um, they accept stablecoins. Stablecoins are now actually, like, 20% of their volume, and their use case for stablecoins is basically, like, very global and very high cost. And so the global means ACH isn't an option, right?

ACH is usually what folks in the US would go to for low cost. If you're gonna use something like, you know, international cards though on, like, a very large basket cost, you're talking about paying four and a half percentage points literally just to international card costs.

And so that's just taking a bunch of your margin that you don't, that you don't wanna give away. So now 20% of the volume comes through stablecoins. We actually did an experiment with them, and half of that is fully incremental, which is to say they would only have 90% of the revenue they had had they not, uh, opened up stablecoins.

The other half is a shift from other payment methods to stables. And then on the cost side of the house, you know, the cost of stables for them is, like, 1.5 percentage points versus 4.5 percentage points, so that's a couple extra percentage points in their pocket, which they don't mind either.

So stablecoins is, is another thing that we're seeing, um, AI companies adopt pretty quickly. I like to say that nerds buy from nerds.

Okay.

And so there's a nice sort of network effect there, right? Like, if people who use AI have stablecoin wallets, when they go to the next AI provider, they have a stablecoin wallet, right? It's sort of self-reinforcing. We also see this with Link, which is our consumer product.

Like, Link just passed 200 million consumers, so it's not a small network. But what I think is more interesting is in the case of AI, it's a very, very dense network. So Lovable-

Yeah

... accepts Link. 58% of Lovable's volume flows through Link. So for every three people who are buying on Lovable, two of them are buying with one click Link checkout because they already have a Link account, and I think that just, like, gives you a flavor of, of the, the density of the Link network, but also the density of the AI network.

Yeah. Is there a... Are there classical measurements of network density that you keep an eye on? You know, 'cause obviously as an economist, like, that's the first thing I go to.

Yeah. Um, I mean, the Herfindahl index is less about network density in particular and more about as we look at all of the transactions that are flowing through Stripe, how concentrated are they on... I mean, you can look at it along a lot of di- dimensions.

You can... And merchants, how concentrated are they on certain merchants versus spread across merchants?

Yeah.

How concentrated are they on certain industries versus spread across industries? How concentrated are they on certain geos versus a broad range of geos? And yeah, we definitely track that concentration. Now, for us, some of that is actually the inverse of network density, which is we want diversification.

Like, you know, um, and we wanna be exposed to many different industries and many different markets and have global reach because the current wave is AI, and I'm incredibly bullish on AI, but we really wanna be growing the GDP of the internet broadly, and that's not constrained to only the AI domain.

Yeah. Um, excellent. Should we, uh, move into the ACP? Uh, I don't know how to transition it better than that . I feel like, I feel like agents do want to eventually do commerce between themselves. I guess that's the transition, and that, that, that, uh, is the perfect intersection of financial infras- infrastructure and AI.

So maybe, uh, could you tell this the story of ACP, right? Like, I, I think, uh, this is one of the, the, the biggest launches of, I guess, like, in the, in, in, in, in the second half of the year a- and, and, like, I guess, a really important strategic move between OpenAI and Stripe.

Yeah. So, you know, we talked a bunch about AI companies in general. One important slice of AI companies is AI commerce, agentic commerce. And, you know, I think just zooming back, like, we're all spending more and more time in some combination of broad consumer-based tools like ChatGPT and AI dev tools like Replit or Vercel or whatever, and we want those agents, those tools, to increasingly take action on our behalf.

And I think, you know, we saw an early version of this in ChatGPT with Operator. But an important area we want them to take action is buying on our behalf. You know, sometimes it's recommending products, but often it's, like, literally getting it all the way over the wall.

So, um, a couple weeks ago, we announced our Agentic Commerce Protocol, which is joint with OpenAI, and it's basically just a shared standard for how businesses can talk to agents. So if you think about it, like, it used to be that a human was buying from a business.

Now, there's an agent that's sitting in the middle, and that fundamentally needs to change how the financial infrastructure works. Like, checkout needs to look different, fraud checks need to look different, payment flows need to look different. But also merchants are trying to figure out how they can-

Mm

... efficiently expose their product catalog, their inventory, their brand, their pricing through a range of agents to have ac-Access to that new stream of demand. And it's kind of a brave new world. So the Agentic Commerce Protocol is really about that shared language for agents to get from merchants what products they have available at what prices, how they want the brand to appear.

Um, and then we also built a shared payment token, which basically allows the agent to pass over the required payment credentials on behalf of the buyer because the agent doesn't wanna bear the risk. The agent doesn't actually wanna be in the middle of the transaction, and the merchant, um, wants to, you know, undertake the charge and actually have the direct relationship with the consumer at the end of the day for returns and more.

And so the shared payment token was also an important component, and then fraud was another important component, right? Is this a good bot or a bad bot? Um, there was a day not very long ago when the optimal thing to do as a business was to block all bots.

Um, now many bots are good bots. You do not want to cut off that demand. And so, um, what we pass over as part of the shared payment token includes scores, um, on the goodness of, of the transaction so that the merchant can, can make the right decision.

Um, one of the ways this manifested was an instant checkout in ChatGPT, which... Have you guys bought anything from this?

Not yet, honestly. Uh, I, I've tr- I've tried. Like, it just recommends things, but I always want to take over the last mile of-

Yeah

... checking out myself, you know?

Yeah, yeah, yeah, yeah, yeah.

I... It's hard for me to, like, hand over control.

Okay. Yeah, and, and, and it... I think they're also still, like, iterating on their recommendations to some extent. I... Last night, my daughter was at a class, and so I took my son out to dinner, and he told me he has a school play, and he is supposed to dress as, um, a Spanish shopkeeper.

And so I tried to search for, like, kids Spanish shopkeeper outfit, and they recommended me a $1,300, like, $1,300 bolero off Etsy.

The play is not that important.

So I didn't buy that one. I... The play is not... I love the child, but the play is not that important. Um, but there's a... But there is a lot of great stuff you can buy. And so in the, in the initial, um, instant checkout launch with ChatGPT, um, you could buy from US-based Etsy sellers.

There's over, uh, one million Shopify merchants coming soon, including some really big ones like Glossier and Diory. Um, this week, Salesforce announced that they're also in. Um, and then, you know, this, this... My favorite is that, like, a week ago, two weeks ago, you could have asked, "Hey, um, are the largest retailers gonna get on board with this or not?"

In the last couple of days, Walmart and Sam's Club have just signed up to also make their inventory purchasable through ChatGPT and, and the Agentic Commerce Protocol, which... Like, I don't think that there is a bigger signal on a big retailer being up for it-

Than-

... than Walmart

... the world's biggest one. Yeah.

Yeah, yeah, yeah. So that's, that's pretty exciting. And then, you know, one of the things that is important to us and to the broader ecosystem about the Agentic Commerce Protocol is it's not about Stripe. So that shared payment token I talked about or the Agentic Commerce Protocol, like, that works no matter who your payments provider is.

We can pass the shared payments token over to any other PSP. You don't have to process on Stripe. Um, it's also not just about OpenAI. And so in the same way that, like, you and I are seeing, um, you know, new, new models come online all the time, we wanna be able to move across models flexibly, there's gonna be sort of new agentic buying experiences coming online all the time, and we wanna make it easy for merchants and kind of like one shot to integrate with all of the agents as they come online, and that's what the ACP, um, really provides 'cause it's a standard protocol versus needing to do custom integrations per agent.

Did you guys see Karpathy's tweet about the... He ba- he basically, like, recreated-

Uh, NanoChat?

Mm-hmm.

Okay.

Yeah. So, like, if he can do that, 8,000 lines of code, less than $100, like, you might think that a lot of companies are gonna roll their own really soon.

Of... Using NanoChat? That would be interesting. I, I, I, uh, haven't made that connection yet. That's, uh-

Let's see. Let's see

... let's see.

But I think this basic premise that, like, it may be a winner take all, but it's not yet clear who the winner is, and by the way, like, I hope for the efficiency of commerce that it isn't a winner take all, therefore, um, many merchants need to actually be having their products sold through many agents, um, is kind of the, the premise of ACP, and we're just delighted to see the early traction, um, and, you know, have just been flooded honestly with both merchants and AI platforms wanting to, wanting to join.

So I think, I think we're on the right path.

Yeah. This is the year of protocols kind of.

Yeah. We, we did an episode with Crunch AI, which does, um, uh, web rewriting-

Geo

... for agents, and they have, uh, Him as a customer. They have Skims. And I, I think every brand is like... I, I mean, for brands it's very easy. It's like, "Hey, I don't really care where it comes from."

Yes.

"If you buy my thing, we're friends," you know? Uh, and I think, like, you take away the part which is, like, the most annoying to, to think about, you know, which is, like, the fraud. I do have a question on, like, the good bots, bad bots when it comes to, like, s- scarce releases-

Yeah

... like tickets for events and-

Yeah

... things like that. I think that's gonna be interesting.

Oh, please tell Ticketmaster. Oh my God.

Exactly. It's like, well, but now if anybody can just have an agent that just goes on the website to buy them, now it's like, how do you do the queue? Because everybody-

Oh, yeah

... gets there instantly, right? So, like-

I won't say the company, but I had dinner the other night with the guy who's the CEO of... You can basically think of it as, like, StubHub for country X. I won't say what country X is. And he was selling, I think it was, like, 3,000 Bad Bunny tickets, and he had 400,000-

Oh my God.

Right

... people come to buy the Bad Bunny tickets, except almost all of those people were actually bots. And this is, this is a great example of what we were talking about earlier, around, like, it's not just about the, the fraudulent dispute.

So the conversation I was having, he was like, "You know, we have scalpers, scalpers who have bots. They, they end up scalping the tickets later. They come and they buy. It doesn't result in a chargeback. Like-" They pay, but they're not the people that we want paying.

And so to our conversation earlier on suspicious transactions, like, there are lots of different types of fraud, and thinking of fraud as just things that result in a fraudulent dispute is actually overly narrow. And he wants to block, you know, all the scalpers, everyone who's, like, you know, enumerating through email addresses in their sign-up-

Mm-hmm

... and, and, and. Um, the other thing he said to me, which was interesting, is he, for various reasons, and some of this is actually, like, the, the, the nuances of how his system is built, um, but I think some of it generalizes, he wants to have those fraud signals before they even get to the checkout page.

Mm.

And so how can we understand the customer independent of them entering their payments credentials? And there are a bunch of ways we can, and we can get better there. Um, his particular reason why is he's got... This is a little mundane, but once the ticket goes into the cart-

Oh, it's-

... it can't be touched for, like, 10 or 15 minutes. And so even though, you know, it's successfully blocked at the actual charge time, it's, like, been held from, from those, those other good customers. So anyway, I, I think there's this, there's a lot of exciting work to be done that's actually, like, increasingly possible, not just because of the scale of the Stripe network, but also because of AI around, um, understanding an expansive set of fraud vectors.

And, you know, if, if you think about traditional deterministic systems, you know, you'd write rules like block, don't block.

Right.

Now you can think about, okay, actually foundation model, text alignment, like, human-readable description of, like, why we're worried about this charge, and then today a human, tomorrow an agent sitting on top of that and decisioning, like, reasoning over the model outputs, um, I think that's the world we'll be in in the next three to six months.

Yeah. I think we need to... We have to be careful about rolling those kinds of things out because people get very upset, and justifiably so, when they are denied something that they, they, they should have by a bot that won't explain itself.

Yes.

Right? There needs to be, like, an appeals process-

Yes

... or, like, some, like, tier two human, like-

For sure

... can I speak to the manager, please?

For sure.

You know?

And-

Like

Yes, for sure. And well, humans make bad calls too.

Yeah, yeah, yeah.

Sometimes they make bad calls at higher rates than LLMs 'cause they can't reason-

Totally

... over as much information, but I agree, definitely appeals process. But then also, like, when we actually look at bad actors, it's like a tiny, tiny share of bad actors accounts for a huge volume-

Right, you're just rooting those out

... of the bad outcomes.

Yeah.

And so what ends up happening is actually good actors have worse experiences, which could mean they don't have access to free trials, or they're gated in how many credits they can use, prepayments, or they're just charged a higher price because they're covering for the cost of the bad guys.

Right.

And I, and I actually think there's an opportunity to, like... You know, I'm from Montana, so sometimes I talk about sheep and goats, although I don't actually know which is good and which is bad. Separate the sheep from the goats in a way that's, like, really good for the good actors.

Why are sheep and goats bad? Like-

No, only one's bad and one's good-

Wait, wait, why, why-

... but I don't know which is which

... can't they both be good?

That was, like, the idiomatic expression I said used growing up, just separate the sheep from the goats. And I'm like, "I don't know which are the sheep-

But they're both good

... which are the goats"

... they're both cute, like.

I think sheeps, sheep are supposed to be cute, but I've always been impressed. Have you guys been to Yellowstone?

Yeah, yeah.

You go into Yellowstone National Park at, like, the Mammoth entrance, and there's this just, like, sheer giant, you know, wall of rock. And there's these mountain goats-

The goats just chilling

... like, exactly. They're, like, scaling it at-

Yeah

... 90 degrees.

Yeah. Like, mountains I get, but, like, dams, I, I don't know why they love dams so much. They- ... it's like, it looks so dangerous, and, like, there's-

So dangerous

... babies just walking on there.

So dangerous, yeah.

Yeah, that's it.

It almost looks like AI-generated, the image.

Yeah, yeah.

But it's real. Yeah.

It's real. It's real. Uh, and, okay, I'm gonna do, like, e- economist corner just because, like, you, you know, you, you clearly still identify as an economist. Uh,

I do identify as an economist. It's very strange.

Yeah, yeah.

Yeah.

Yeah. Um, well, so, like, when you encounter those, like, uh, the StubHub for X, don't you feel the temptation to recommend an auction? Like, and there's so many auction m- mechanisms that can clear the market. This is clearly a market, uh, it's called a clearing problem.

What's the market solution here?

I don't know that I am usually tempted to recommend an auction. I am usually tempted to ask a lot of very probing questions about why they've designed-

Yeah

... the system the way they've designed it, and then try to brainstorm, yeah, whether there's a, whether there's a more efficient path. But yeah, I, I feel that way about, um, most, most pricing, matching, discovery, recommendations. Like, most markets are just inefficient, and so I think there's a, a lot of opportunity to make it better.

Um, one of the reasons I joined Stripe, and one of the things I have loved about being s- at Stripe for the last four years, is when we see those opportunities to make the market more efficient, um, we can actually invest in doing them without optimizing them, you know, without monetizing them directly.

So you can think of it like incentives are very aligned. Anything we do to help the businesses on Stripe grow helps us grow because, you know, they-

Yeah, you-

... they run their payments-

You take a percentage

... through us.

Yeah.

Um, and so-

It, it's like-

... we do this all the time. Like, here's how to, you know, improve your checkout, and we just, like, update the checkout for them. Or, like, optimizing their payments acceptance or automating their retries. Um, so anyway, I, I think it's just, it's, it's very nice to be at a company where you don't have to worry about the go-to-market for something that helps the businesses that run on you.

You just have to help the businesses that run on you do better. Um, and that in and of itself is, is a good outcome-

Yeah, 'cause they're very-

... for your company and justifies the investment.

Yeah, they're very incentivized. It's like top line and, and bottom line sometimes.

Yeah. And by the way, like, this isn't all causal, but, um, last year the companies on Stripe grew seven times faster than the S&P 500. And, you know, again, like not all causal and there's selection effects, but definitely-

It just-

... some of it

... it's just for sizing effects.

Some good, some good tailwinds. Yeah.

Yeah, so I mean, yeah, and, and I think The Economist term I really like is deadweight lost.

Yeah.

And, and, like, basically, like eliminating friction, it inc- it, it means improving surplus for both producer and consumer, and Stripe benefits, uh, as a result, which is, like, that's nice. Everyone wins. Uh, it's a, it's a good, good, uh, fuzzy feeling.

Coming back to the protocol, I think that it's an interesting decision to actually release it as a protocol. Like you said, it's gonna be many to many, and like sometimes Stripe is not involved. You also mentioned, like Stripe Link and Stripe Checkout, and those are Stripe products, right?

Those are, those are not protocols. So I think, like, it's a very interesting and pivotal decision to choose to release it as a protocol, as, as opposed to not. I was wondering if there's, like, any internal debate, or is there any internal color about, like, the decision behind choosing-

We didn't debate it

... a protocol. Yeah.

Um, you know, Stripe has moved fast for the entire four years that I have been there. I think it is accelerating, and I think it is accelerating because customers, users, changes in what the market needs, both what businesses need and what buyers need, um, in the world of AI is accelerating.

And so new products, new solutions like ACP, like token billing, are literally being pulled out of us, and what we saw was a hole in the market. Like, consumers... And we can talk about developers, too, but, like, consumers wanna be buying through agents.

Agents are ready to buy for them. Merchants are ready to let agents buy on behalf of consumers, and yet the market can't figure out how to make it work, and that's not about Stripe. That's about growing the GDP of the internet.

That's about making sure commerce can flow. And, you know, so, so there was no debate. It's like the ecosystem needs this. Like, we will pair with OpenAI to get it over the wall, and, and by the way, like, as I said, we're early days.

You know, ACP is seeing a ton of traction, but if it wasn't or if something different comes out six months from now, like all we want is a shared standard. We don't need to have, like, our name on the shared standard.

I think it's absolutely, absolutely the right thing, the right thing to be doing. You know, I have had folks ask me, like, "Is agentic commerce just a straight substitute for the commerce that's already happening on the internet?"

Yeah, it's just like fancy APIs.

Exactly, and I think the answer is it's not a straight substitute. I think it is actually, like, expanding the aperture of what commerce will get done, and the first person to put this bug in my head was Dwarkesh.

Mm-hmm.

And when he said it, it actually, like, took me a second. Like, I didn't believe it, but I think it's right, which is if you look at the share of income that is spent on consumption as a function of how much income you have, you see that high-income people spend a much lower share of their income.

Low-income people spend a much higher share of their income. There are many reasons for that, but one reason is the biggest cost to very high-income people consuming is not the dollar cost. It's the time cost of consumption. And so I'm very interested in how agentic commerce can open the aperture-

Ah

... for spending by high-income people because it's removing the most costly or binding constraint, which was their time, and we're actually then truly pumping incremental, not substituted, but additive dollars into the economy and obviously, like, you know, first, second, third-order effects of that.

Yeah. Well, it results in sometimes buying $1,300 costumes and ah.

I didn't buy that $1,300 costume. Trust me, I would rather spend an hour than buy a $1,300 costume.

Well, you know, just work a few more years at Stripe. You'll get it. Uh, uh, so, so I think, one, like there's some, I think, uh, interesting... Well, I love protocols. Uh, you know, as a developer tools person, uh, I've been, I've been involved in designing a few of them, debating a few of them.

Yeah.

What are the forks in the road that, you know, like someone else was, like, discussing, uh, uh, something really strongly, and we decided against it, or maybe it's still an open question. Uh, I'll g- I'll f- I'll give you one, and then maybe you can volunteer another.

Uh, so the... I, I mentioned that, uh, both, uh, Solana and, well, Circle-

Mm-hmm

... have, have sponsored my conferences before, and they're, they're also trying to be, uh, build a protocol for, for agents. And both of them actually give agents a wallet-

Mm-hmm

... right, as opposed to a payment token.

Yeah.

And I, I think, like, having an agent with their own bank account effectively is an interesting choice. You didn't go for that. Uh, so, like, is that a decision factor, or is, or is there a different one that you wanna focus on?

Let me parse two things. When I think of the commerce protocol, that's primarily around what's the standardized way that businesses expose their products and their inventory and their prices and their brands and make those available to agents to expose to the consumers and/or to buy on behalf of the consumers.

Um, that really comes down to, like, how should a product catalog be expressed? How should prices be expressed? I think the current version is, like, the bare bones version, and it will continue to evolve. For example, like, you could imagine, like, from a market clearing perspective that the merchant should also be, as part of that, articulating the cut that they're willing to give-

Yeah

... to the agent, right? Like, like a little bit-

You're... Yeah, in a, in a classic, like if, uh, the human agent, I give them a budget, like, uh, what their negotiation target is and what their max spend could be.

Yeah. Yeah. Exactly. And so in this case, I'd, I'd be like, "Okay, well," you know, whatever. The, the bolero that I didn't buy- ... was a terrible recommendation, but they have many good recommendations, but that was a terrible one.

Um, you know, this costs $1,300 on Etsy, but, you know, I'm willing to give X to any agent who facilitates the transaction, um, either on top of or underneath, um, the, the 1,300. So anyway, I think there's gonna be an evolution of, like, the various parameters that should be included, but, like, the, the basic set was what do you have to be able to deterministically expose to an agent So that they understand what's available, and what representation of the product and brand and, you know, sometimes it's size, and number, and whatever, like, has to be made available to the agent and/or the, the h- the human that's, um, initiating in the first place.

The shared payment token is a little bit different, which is like, okay, how do you actually get the transaction done? Like, how does the money flow? And even at Stripe, like, that has been evolving. So shared payment token is, is what we built and launched and have in the background of the Instant Checkout implementation with, uh, ChatGPT.

But a year ago, Perplexity launched a travel search and booking agent. Did you guys see this?

Yep.

That is also powered by Stripe, and there the payment flows are a little bit different. So we have an issuing product which allows you to issue virtual cards, and what happens, um, in that sort of flow is the agent gets issued a one-time use virtual card to spend on your behalf.

And, you know, people get very, get very jumpy about that. But I like to remind them that, like- ... when I order from DoorDash my Philz Coffee, right, DoorDash is issuing a one-time use virtual card to the driver for $6, believe it or not, um, to spend, uh, on my behalf there.

And so now you're just inserting, you know, AI agent instead of human agent, and in the same way that my DoorDash driver never saw my card credentials and couldn't spend, you know, more than $6 and had to spend it in a constrained time window near my house, same thing for the AI agent in the Perplexity travel search and booking agent.

Um, and we still have agents doing commerce through Stripe using the virtual card implementation, and there are pros and cons. So, um, I don't think that, like, it's gonna be all virtual cards, it's gonna be all shared payment tokens, it's gonna be all agent wallets.

I think stable coins will be an interesting direction. I think wallets in general will be an interesting direction. I think stored, stored balances will be interesting, especially as we're talking about kind of micro-transactions, right? So we're talking a lot about buying, like, goods.

Buying goods are usually priced high enough that it's worth sort of like a card transaction type approach, but if you're talking about buying AI or buying some-

Content

... inference or buying content, you wanna be able to make 5, 10, 25, 50 cent transactions, and those are hyper inefficient in the card world. So I think agent-to-agent payments are gonna push us a little bit to the next frontier here.

Um, but ACP, again, is like distinct from how the shared payment token works or how the money flows, and I think that will, A, continue to evolve just because the market needs are evolving and the technology possibilities are evolving, um, but B, will also, like, also doesn't need to be s- standardized in the same way.

What about the receive side? I guess, can my agent make money for me?

Can your agent make money for you?

Yeah.

Oh, I was thinking the opposite, which is like, I'd be happy to give an agent, you know, $3 to go out and do deep research for me, and so we're, we're trying to figure out how to enable that.

Oh, no, that's research. No, I, I was like re- you know, like, uh, spending I think, like, the, we have a good mental model-

Yeah

... of how to spend.

Yeah.

Especially 'cause we have human agents as well helping us to spend.

Yeah.

Uh, making money is, is, is obviously the, the original draw of, of, of, uh, Stripe for any, any founder.

Yeah.

Uh, yeah, I'm just kind of curious what's, what's your thinking there.

Well, so we make it easy to monetize your MCP server, so that's like one thing. We also are seeing an increasing number of new businesses get going in AI dev tools, like Replit and Vercel, and so we want to make it really easy to spin up monetization also within those tools and within that flow, not be taken out of flow and go and create a Stripe account, authenticate, and whatever.

Um, a couple weeks ago, we released claimable sandboxes. Have you guys seen... been in Vercel and seen it or any-

I-

No

... I know they have a sandbox product-

Okay

... but I don't know about claimable.

So we have a sandbox product which you've seen from, like, being within the developer experience on Stripe. Now, that sandbox product can be, like, invoked, used, manipulated. You can create your products and your pricing and run test charges and generate customers and in that sandbox before you even have a Stripe account, or maybe you have a Stripe account from your last business, but before you've linked it to this business, right?

And, and we call them claimable sandboxes, and Vercel and Replit were... I actually remember talking to Replit at Stripe Sessions in May about our sandbox product, and they were explaining to me how, um, you know, people are trying to build businesses end to end, and, like, one of the wonky parts of the flow is setting up the payments integration and get going.

And so we started talking about, like, okay, could you actually have, like, the sandbox environment there? Um, and it was, whatever, four months later and they had it. They launched it. We launched it with them, and also, um, with Vercel.

Actually, Guillermo had a cool post a couple days ago that, um, our Stripe integration is now one of their top, I think it's, like, the third highest, uh, integration that they're seeing in v0. Um, and we're only two weeks into the launch.

Oh.

But it's basically like all of these people aren't going to create v0 just for fun, to, to create something, just a website or whatever, just for fun. They're going to create a business, and so making it really easy for them to do the business, payments-

Yeah

... backend part of that. And how it actually works is literally like you're right there in v0, and you, you know, you have your plant shop, and you create your products- ... and you set your prices, and you run your test charges, and you click a button at the end.

Like, when you like what you see, you can go back and iterate later, but you click a button, and your tab opens in Stripe, and you can either sign into the account you have or create a new account, and you claim that sandbox.

And that sandbox, you can take it live, and it becomes your business. And lots of people are taking it live every single day, and we're seeing new businesses get created, um, that were never before. And one of the things that's really fun to see, I was going through the list of businesses, I probably can't name them live, but, like- A chunk of them are AI companies.

Like, a chunk of the startups being created today are AI companies, but a chunk of them are, like, non-technical founders who may have actually struggled to, like, get going on Stripe had they not had it kind of within that v0 experience.

Yeah, low code is a huge enabler.

Low code is a huge enabler, and, you know, we've, we've done some good work in our own onboarding experience to make it low code, and we have, you know, low-code subscription and whatever. Um, we have our, our new onboarding experience.

Internally, we call it future onboarding experience, and it kind of walks you through what's the business model you're trying to create and then sort of, um, stands up the sandboxes for you. But what's cool is now you do a bunch of that in, say, v0, if you're in v0, and then you come over and, and your future onboarding experience has already learned, like, your intent and your preferences and all of that-

Oh

... from Vercel, and so you're dropped in X percent of the way through, uh, with the sandbox already spun up.

So this is, like, from the outside how people perceive AI at Stripe. What about inside? So you mentioned 3.5 was kind of like the moment you took it seriously. What were the first internal use cases, and then how do you use AI at Stripe today?

Yeah. First internal use cases were, you know, bottoms-up experimentation, right? So we created... We call it Go LLM, but it's like just a ChatGPT-like interface where you can engage with a bunch of different models. Um, it was the very, very first version actually wasn't like an LLM proxy where you could build production-grade systems.

It was literally just like ChatGPT-like stuff. And then we had this, like, preset feature, which was, like, prompt saving and sharing, and so you could share your temp- oh, you know, this is how I figure out what customers to reach out to and generate reach outs or, um, you know, rewrite my marketing content in Stripe tone or whatever.

And you had sort of hundreds of presets that came on, like, overnight 'cause, uh, everyone was into it. And then we generated... So then LLM Proxy was like, okay, now production-grade access for engineers to these LLMs, and a lot of the early use cases there were actually around merchant understanding.

So I mentioned a little bit ago, but we have like thousands of merchants that come onto Stripe every day, and we have to understand: Who are they? What are they selling? Is it supportable through the card networks? Like, are they credit worthy?

Are they fraudulent? Um, and there's a lot that LLMs can do there. So those were some of our earlier, um, earlier use cases. Fast-forward to today, I mean, you know, I, I actually, uh, was looking at the dashboard earlier 'cause I was planning for, for 2026 and some of our LLM costs.

8,500 Stripes a day use our LLM-based tools. Okay, there's like only 10,000 Stripes. Like, not everyone is in every day. Like, it's basically, it's basically everyone. Um, and you know, I think people are getting, like, pretty creative in, in the applications.

Um, I was talking last week to the LPM team, so local payment methods. Um, you know, you and I think a lot about cards or whatever, but local payment methods matter because the businesses on Stripe are almost always selling internationally.

And when you're in other countries, having local payment methods, you know, GiroPay if you're in, um-

Yeah

... Germany.

I'm from Southeast Asia. Yeah, it's all over.

Yeah. Wait, what's your favorite?

Uh, well, no, I mean, uh, there's like GrabPay, I guess.

Yeah.

I don't know. Yeah.

Yeah, exactly. And like, if you don't see a local payment method, like, if you only see cards, you might not have cards, or if you only see cards you might not feel like it's like localized to you or meant for you, whereas if you see like in-market regional payment methods, you feel much more connected.

You're much more likely to convert statistically, and oftentimes the fees also, uh, make more sense, um, for, for the merchant. So we've invested a bunch in integrating local payment methods. Most businesses on Stripe use our optimized checkout suite.

Our optimized checkout suite comes with over 100 payment methods out of the box. But one of the most requested features we get is payment method X, payment method Y, payment method Z because I also wanna be in country HK.

And so, you know, what is the, the challenge? The challenge is integrating with any new payment method, and it takes like two months for a couple engineers. It's not the end of the world, but Stripe's a pretty lean company, and we got a lot of stuff to do, and so, you know, for the marginal payment method, is it worth it?

Yes or no? Well, when you step back, what are you doing? You're really, like, looking at Stripe's code base. You're looking at, like, how the LPM works and, like, the integration guide for the LPM, and you're kind of like hooking the two up.

And so the LPM team, it took them two weeks for the first one, uh, but they just launched a new Pan-European payment method in two weeks using an LLM to, like, build that integration, and I think they'll probably, you know, have it down to a day or two within, within a month.

And so that's just a good example of, like, it kind of should be just like a machine talking to a machine, and there's pretty good documentation on both sides of the house, and so the LLM can make it, can make it pretty far.

Um, we also use a lot of AI coding assistants, and, you know, I would say like 65, 70% of engineers use them on the day-to-day. Um, I have a really hard time understanding impact. Like, I don't quite know what statistic-

Mm-hmm

... to look at. I, I don't actually... I think-

It's not lines of code. It's-

I don't think it's lines of code-

... number of PRs

... because-

Yeah

... I have in the last week been sent three different documents that I know were like at least partially written by- ... an LLM. Documents, not code. And in all cases, I went back to the individual and I said like, "I actually just wanna see the bullets that you put in ChatGPT or whatever."

"Not the, not the eight-page document because I have a really hard time reasoning about the eight-page document is... And it sounds good, but I'm not quite sure it's like connected to reality." And I feel the same way about lines of code.

Like, I don't really want more lines of code, just like I don't want more pages of docs.

Yeah, yeah.

So we're watching that. And then also the cost of a lot of these coding models is actually like pretty non-trivial.

Mm-hmm.

Um, and so as we're planning forward to next year, we're reasoning a bunch about like where can we get somewhat more efficiency there given like obviously it's valuable and we want people to, um, be using AI coding tools for sure, but and we wanna make sure that, that we're getting the right returns for the business, and some of that is managing costs and some of that is getting a clearer read on impact and value.

How do you feel the social, social contract is changing? Like you mentioned, just send me the bullet points, right? It's like I could've sent you the bullet points before LLMs, but you are making me write this memo, right?

I feel like in a lot of organization there's like a performative... And part of it is like, you know, wearing a suit to an important meeting. It's like in a way it's like, "Hey, I'm doing it-

I don't wear suits. I don't wear suits. I hear some people do. Yeah

... you know, I'm doing it to show you respect."

Yeah, yeah, yeah, yeah.

And in the same way I could've just sent you this bullet points-

Yeah

... or we could've had this meeting in shorts.

Yeah, yeah, yeah.

Do you feel like with AI now it's like, okay, if you're gonna do it with AI, just send me the bullet points, and we're kinda like breaking through in a way of like-

Yeah. So that's really interesting. Okay. So I hadn't thought about this before, but here's my working hypothesis. Tell me if it tracks. What I care about is that the expert in the area, like they're an expert in the area, otherwise they wouldn't be sending me a doc, right?

The expert in the area has thought deeply. And what is writing, like actual writing, typing, whatever, doesn't matter, but like writing not with an LLM force you to do? It forces you to think deeply. It forces you to structure your reasoning.

I don't know about you guys, but when I read a doc-- when I write a doc, I've like read the doc like 50 times and thought about like, "Does this logic track? Are there gotchas I'm not considering? Like, is that the right train of thought?

Like, how might somebody else look at this?" And it's not like that I wrote the doc to be performative and the bullets would've been better. It's that the careful, thoughtful, arduous, time-consuming construction of the doc forced me to appropriately reason from first principles.

And I think LLMs do the opposite. Like, oh, you just throw in the bullets. You don't have to reason from first principles and it sounds good and people like... But I think that's extremely dangerous.

I think that is true, but I feel like you still are not generating documents with AI. So because you are the type of person that uses the writing as the thinking-

Yeah

... you still go through the process.

Yeah.

Versus the people that use the AI still wouldn't have put that much thought into writing the long document anyway. I think to me that's really the thing. Same way we were talking about this, uh, for code yesterday-

Yes

... in another interview-

Yes

... which is like the slot machine effect of like Cursor and like these tools. But at the end of the day you gotta merge the PR. So you gotta come up with something that makes sense for the business.

Yep.

Like with these documents it's kind of the same, right? It doesn't-- Like you just need to come up with the right ACP design. I don't care if it's 10 bullet points or like 10 pages.

Yes.

To me I think like things are changing now because also people read more summaries. So it's like, well, if you summarize my thing, then why should I write a long thing?

Yeah.

I should just write a short one, so.

Yeah.

I, I mean, I don't have an answer, just like interesting to see how, you know, you're like, "Just send me the bullet points."

Yeah. The primary thing that I... Well, there's many things I care about, but like a very concrete non-negotiable is if an LLM was used in the generation of this content, please cite the LLM.

Mm.

Because my least favorite thing is to be like two pages into what I think is a thoughtful doc and find the annoying like space, double dash, space, and then-

As a double dash guy, I feel like- I mean, I was writing it before LLMs ruined it.

No, that's not what tips me off. That's not the only thing that tips me off. But y- you know what I mean? Like I g- I g-

Yeah, yeah

... like I do think we all need to be careful about, um, LLM slop. And then I think there's like a societal behavioral thing here too, which is like we can't turn our brains off. I mean, I actually think that like what do LLMs make all the more important in the world?

The ability to think and reason deeply. To like question, to like tell the machine what to do more so than like to do and execute the task, 'cause the LLM can do and execute the task. And so if you're, if you're looking at LLMs and you're like, "Oh, that makes, that means I don't have to think deeply 'cause they're just gonna do it for me," which is very natural 'cause they do produce enticing output.

Mm-hmm.

I just think it's like, I think it's like risky for, for society. I mean, I think we've seen how like who- people who grew up on social media have like a lot of issues with attention. I think people who...

I don't mean attention trying to get attention.

Yeah, yeah, yeah.

I mean attention like staying focused on a task. I think in the same way, people who over grow up in their work life on LLMs risk under investing in depth, and I think that's particularly dangerous in a world where with LLMs actually it may not appear this way in the moment, but like depth is the most important thing.

I would push back a little bit in terms of... I think I may be a bit more slop-friendly than you guys.

Are you?

Uh, be- just because like, uh, hu- slop comes from humans and slop comes from AI.

Yep.

What just matters is like when you, um, sign off, when I send you the document or when I send the PR, I am signing off on the whole thing.

Yeah.

I, I can't abdicate responsibility to the LLM.

Yeah.

Maybe LLM had good output, maybe not, but I'm the, uh, final judge. I'm the editor, right? A- and so like, uh, I don't-

Well, so I actually think we're on the same page though. So I am like all for using it in the generative process.

Right. Yeah.

Actually it was really cool when we were-

It's a tool for thought

... it's a tool for thought.

Yeah.

And it's a tool for rapid experimentation and rapid iteration, and I love to look at like a demo or a prototype that like I don't wanna see a doc on it. I wanna see like the quick thing that you spun in whatever tool you use.

Actually, when we were working on claimable sandboxes, I'll never forget Vercel sent us basically like the V0 of like how they thought it should be implemented. Just like the UI. Like this is what we think the experience should look like.

And it was extremely clarifying. Like much more clarifying than hours of meetings and pages of design docs. And so all day long on the generative, but like you need to like deeply put your stamp of approval before you push the PR, before you publish.

I did wanna double-click a little bit on both, uh, you know, the like two, two primary use cases on, uh, RAG and writing code. Just on, just I guess on like in- internal information-

Yeah

... there is obviously Glean, which we talked about yesterday, and just all the other, uh- ... internal code search tools. You guys use Notion as well. Is RAG still relevant? Is that something that's, like, in active development, or, uh, what, what's, what's beyond that?

What, what's, like-

Yeah

... the frontier?

So we've actually been leaning in really hard on, we call it Tool Shed, but it's like a, it's like an internal, like, MCP server that basically has access-

Everything

... to, like, all the Stripe tools.

Yeah.

And, you know, what I like about that is, like, it's, and it's managed centrally. Like, you know, there's, you know, it's managed centrally, and it plugs into... We've since killed that GoLLM thing I talked about, and we did a new, like, a, you know, implementation, like, open source, like, LibreChat situation, which is great.

But it, like, hooks up to all those same Tool Shed, like, you know, MCP servers, and it's got, you know, I mean, everything you would think, like Slack and Google Drive and Git, but also, like, access to Hubble.

So it can see, like, our data catalog and all the data, and it can query the data and, and, and. And I think that's been really powerful. I don't think RAG is dead. I do think there's an important name of the game around it's not just, like, the information that's available in all those tools.

It's also being able to interact with all those tools, right? It's like the tool calling, and so I think they coexist, and I think they, I think they coexist together. Also, while Tool Shed is owned centrally, anybody can a- so you can a- you know, the, the Salesforce team can add Salesforce and, and, and because we don't want to be blocked on some central team in order to have those tools exposed to, to the LLMs and to the agents for Stripes.

Yeah, you want to decentralize a bit. Uh, and then code-wise, uh, you know, coming on the code side, uh, obviously closer to home for me, I think it's also, like, an economist's, uh, problem of measuring productivity.

It is. Okay. Well, now you're just making me feel guilty for not having cracked it.

No, no. I mean, uh, it's, it's unsolved globally.

I'm kidding. I'm kidding.

So, like, how-

Yeah, it's hard. It's hard.

Yeah, and, and that's the thing. Like, you, you, you're looking at the cost, and you're like, "Oh, it's pretty high. I don't know. Maybe we, like, move to open source models or something." But, like, you don't know the productivity gains you're als- you're, you're getting in, in, in, in interim, and you have pretty expensive engineers.

Like, it's, uh, it's hard to tell.

Engineers are expensive. Um, engineers also, like all of us, right, are hyper-motivated to do the best work of their lives, and so there's an important component of, like, if the people want it, like, there's inherent value in, in providing it, right?

Like, when people have the tools that they want, are learning the things that they want, like, they work harder. They're more creative. They produce better output. So anyway, there's all this, like, sort of soft fringy stuff that I think is added value above and beyond the actual, you know, production output, and then there's also, like, the learning curve, right?

So I mean, we think about this a lot, like, when you launch a new traditional ML model or now, like, AI solution, right? Like, if you over-focus on the results in the immediate short-term window, you really risk getting a false negative.

Like, it's not good enough yet. It's not tuned yet.

Mm-hmm.

It doesn't have the feedback loop yet. You know, hasn't had time to get the training data to get better, and I think that's, like, kind of particularly true in, you know, AI because you, when you're working with LLMs, it's like, well, with, like, GPT-4, like, it didn't work, and then you swap in GPT-5, and all of a sudden it does.

Or we use, like, uh, GPT-4o, and it was, like, kind of a little bit expensive to justify the humans that it was replacing for a particular risk-related task. But then next thing we know, like, o3-mini is out, and it's like, you know, $3 million a year savings for the business, both because the model is less expensive, but also more importantly because it replaces more of the humans.

And so I, I do think that, like, when I think about the optimal adoption of these AI tools, it seems wrong to focus on in-year ROI, and it seems right to focus on two-year, three-year ROI. Now, inherently hard to know what two or three years is gonna look like, but if we look at the history, models getting much better much more quickly, models getting quite a bit cheaper quite quickly.

Um, it overall makes me bullish that we shouldn't over-obsess, um, around, around in-year returns.

Yeah. In-year return side. That's, that's a good term I never thought about.

Especially when the economy isn't doing so well.

Amortizing it like that.

Yeah.

Yeah. Yeah.

Um, what about data?

What about data?

You know, it's kind of like-

Yes. I was gonna move to the data side.

Wait, what about data?

Uh, yeah, you know, it's like, oh-

All right. Haven't been talking about data

... are the engineers more productive or not? Like, uh, what about... Yeah, like, I mean, text to SQL, right?

Yes. Yeah, exactly.

It's kinda like the first iteration of this.

Yeah.

Like, what's the productivity like on, like... I mean, you can generate any chart-

Yes

... now, right?

Totally.

But, like, doesn't mean it's good.

So yeah, yeah. So we have this... He's not really a guy. He's an AI, but his name's Hubert, which sounds like a guy. So we have this guy called Hubert, which basically is, like, natural language to ask questions about the business.

By the way, we have a Sigma assistant, so, like, our users can query stuff about their business on top of Stripe data. That's a much more constrained problem because your Stripe data is, like, your revenue data.

Yeah.

It's, like, very well structured. It's very well documented. It's available in the dashboard and in Sigma and in Stripe data pipeline. You can hook it up to whatever. And so, you know, a text to SQL experience sitting on top of that, like, it's not gonna be perfect, but, like, it's pretty airtight.

And by the way, like, if you use natural language to describe what you want, we write the query, and then we also tell you in natural language what the query does. So even if you're non-technical, you can validate it.

Okay, now imagine there's a lot of tables at Stripe. There's a lot of nuance in Stripe data. There's a lot of nuance in Stripe's business model. Um, Hubert is the guy that sits on top of this Hubble tool, which you use to find, explore, and query Stripe data, that does that internally, and it's early.

I mean, we have, like, 900 people who use it a week. We have tried to focus the people who use it mostly on technical folks who know the domain for exactly the reason you were citing earlier, which is it could get the answer fundamentally wrong, and technical folks are gonna be better positioned to validate and provide feedback.

One of the most interesting things to me as I was going through the Hubert evals was the place where Hubert did the worst was around data discovery That is to say, it had a hard time finding which table and which field was best to answer the question at hand.

Mm-hmm.

And, I mean, personally, as a user, when I know the table, I actually now just articulate the table in the, in the chat interface. But more importantly, we're doing a big push right now to deprecate low-quality tables and have the owners document high-quality tables.

I haven't yet figured out if I trust an LLM to do the documentation, so for, like, the canonical data sets, we're kind of brute forcing it with, with humans who, who know the domain. Um, the other thing that we are exploring but we haven't landed yet is offline it looks like there's actually really in...

Like, Hubert does much better if you tell it where in the organization I sit because it knows, oh, I'm interested in LPM data, or I'm interested in link data, or I'm interested in OCS data. People who work on the optimized checkout suite tend to query these tables, look at these fields, ask these kinds of questions, look at these types of metrics.

Now, that's not in production, but I, I think there's this interesting question of, like, we can have humans do some, like, prompt engineering or documentation or whatever, but we can also give the LLM more just historic context about what people like me liked to do, basically.

So that's, that's the next step there. I'm bullish on it. I think... I don't know if it's gonna be two months or two quarters before, like, everyone's on it. Like, I, I think it might take some time to get high enough conviction that we're not gonna have an important wrong answer.

Mm-hmm.

Text-to-sequel is really easy, though, with really well-structured, well-documented data.

Yeah, yeah, yeah.

It's just most data is not-

Well sequel, yeah

... well documented and well structured.

Well, so, uh, before... Immediately before this, I was actually in the data engineering industry. We talked about DBT5Tran, and you said you didn't have an opinion. But I always thought that, like, data discovery is the im- important victory or, like, the, the, the, the ultimate victory of data catalog people and semantic layer people.

Do you agree with those movements in the data world? Do you have any tweaks on the modern data stack that you have-

Yeah

... opinion?

So we are increasingly moving to, like, semantic events infrastructure. I think the value of near real-time, high-quality, well-documented data is about to skyrocket because I'm pretty sure that nine months from now, no one is gonna wanna go and, like, look at a even, like, static dashboard and click around.

They're gonna wanna be fed insights, or they're gonna want their agents to be fed insights, and they're gonna wanna be able to just, like, pull real-time high-quality data. For us, what that looks like, like the, the two most important domains for our users in that regard are payments and this, like, usage-based billing, which needs to be very, very real time for all sort of, like, the AI business model stuff we talked about.

And so there our, our path is, like, semantic events, canonical data sets available in near real time in dashboard, yes, 'cause some people will still use dashboard, in Sigma, so, like, queryable, um, but also in sort of a, a Stripe data pipeline type, right?

'Cause very few people wanna look at Stripe data in isolation. They wanna look at Stripe data connected to the other stuff, right? So I mean, you could just even imagine, like, pulling into BigQuery or whatever. They wanna see it connected to other stuff, and, you know, historically, honestly, like, it wasn't all the same data feed for all of those products, and that also creates confusion.

So we're doing a bit of a, of a rearchitecture for that flow starting in the next six months, which is billing and payments, and then I, I think we'll, I think we'll expand from there. There's always gonna be a bunch of data that for whatever reason doesn't fit your, I like to call it a North Star architecture, but, like, your North Star data architecture, right?

And I wish that someone could figure out how to make sense of the old bad data-

Mm-hmm

... so you don't have to, like, rearchitect everything and throw away the old. Right? Like, you'll always have, like, okay, you know, like, there's, like, the stripe.com website, which happens, like, you know, before you even create an account.

It's a very different type of data. But, like, you know, how, how should I reason about that? How should I manage that? And I, I think traditional enterprises deal with this a lot and then-

Like converting website analytics to signed up users and all that?

Yeah.

Oh, I've thought about that so much.

Oh, yeah?

You just need, like, uh, it's kind of fingerprinting, which is, like, something that people are kind of, um, uncomfortable with, but you can. You can do it.

Yeah, yeah, yeah.

We, we have, like, a couple questions on, like... I think there's a, there's a build versus buy question-

Yep

... on you're building a lot of internal tooling, and that's great, but also there's a lot of great tooling out there. How do you navigate this? Obviously, you have a lot of unique internal context, but you also have...

You work with external vendors. Like, people, I, I guess, want to know how to work with you, but also people in your shoes at peer companies also want to know how you d- do this decision.

Yeah. I think for us it's not an either/or. It's very much an and when it comes to build versus buy, and some of that and is sequential, right? So you, you and I were talking earlier about, like, when, when GPT-3.5, I think, like, first hit the scene, we were like, "Oh, everybody at Stripe needs to have access to LLMs," but, like, we don't quite know how to do that in a way that's, like, enterprise grade, safe, and we feel good about.

Like, we don't see a provider there right now, and so we built it. But now we use, like, open source, Libra Chat, you know. So I think it... there's, like, there's, like, an evolution over time, and one of the things that I think can be really hard, especially for the team who has tunnel vision for the products they own, you know, you love your product, you wanna make it better over time, is you can get stuck in a lot of hill climbing.

Like, we could have taken GoLLM and been like, "Oh, we should figure out a way to, like, give it access to Tool Shed. Oh, we should figure out a way to make it do, like, deep research," or, you know.

Like, we, we, we could have done that. Um, and sometimes you just need, like, sort of more of an outside in perspective of, hey, if I ignore the sunk cost fallacy, ignore my emotional connection to the thing that I spent nights and weekends building- First principles, like if I were to do this today, what would I do?

And some of that is also making sure people feel a lot of confidence and conviction in their own abilities, and the fact that there's a ton that they can contribute to the company across domains to kinda liberate them from, from, from needing to own, needing to own this product.

Um, another thing that we've done, and this is especially true in the AI space just because, um, there's, like, so much new stuff coming online. We call it the Spotlight program, but basically, like, we put out RFPs for products that we wanna buy.

Ooh.

So one example that we did recently, I guess it was like a year ago now, was evals.

Mm-hmm.

So we were like, "Okay, here's our problem with evals, like, who's gonna solve it?" And we put out this RFP in the Spotlight program. We obviously see a lot of these companies directly because they, they run on Stripe and/or, like, their investors have some affiliation with Stripe, and so we know them.

So we actually had, like, more than two dozen applicants for this evals RFP.

There's no way you can evaluate all of them.

Well, we... Actually, we did. So the, the-

Oh my God

... they, they wrote, like, nice one-pagers. We read them all. We narrowed it down to two finalists. BrainTrust ended up winning. We did a POC with them.

We've had them on, yeah.

They rock. We stuck with them. We also love, like, Weights & Biases, um, Flyte, Kubernetes, like sort of like, sort of like the basic stack you can think of. But there've also been cases where we have had to build.

So one example is if you think about traditional ML for a second, also relevant in the world of, like, our foundation models in that embeddings basically, like, can become, uh, features, our feature engineering platform. So we had a homegrown feature engineering platform.

It was old. It was on its last legs. Um, we had a team internally that, um, really wanted to adopt Tecton.

Mm.

We evaluated Tecton. This was a couple years ago now. At the time, we couldn't wrap our heads around using it on the charge path, just from like a latency and reliability perspective. Like, we've gotta be operating at six nines.

We gotta be, like, decisioning in, you know, tens of milliseconds for some of these models. Like, we just... we, we can't reason about that on the charge path. We ended up pairing up with Airbnb and, and building, um, w- we call it Shepherd internally, but it's open source under the name Kronon.

You know, I think there are, there are cases, there are cases for both. But we always start with, like, what could we buy? Sometimes those are obvious solutions. Then we say, "Oh, there's no obvious buy solution, but, like, maybe there's some new startup thing.

Let's run the Spotlight program." And if we really come up dry, then we will build. And as we build, we reason about, okay, six months from now, 12 months from now, should we still be building-

Mm

... or should we actually swap in, uh, a buy solution 'cause the market has evolved? Um, the other thing that we feel pretty strongly about in the world of AI is there will be many model providers, there will be many models, and we do not want to hitch our wagons to just one horse.

You know, there's lots of enterprise-grade versions of choose your LLM provider. Like, that's of much less interest to us, um, than solutions that sit on top of many different providers and many different models and allow us to swap in and out.

Yeah. And this, this, the Stripe Experiments team-

Yeah

... when you decide something we need to go-

Yeah

... you have the same people always do this, or do you rebuild this team based on if it's evals or if it's, like, something else?

It's-

It's probably bottoms up, like whoever-

Yeah

... how-

It's, it's bottoms up. So experimentation happens in a lot of different places. So, like, for the evals, we just did it, like we ran it within ML Infra. With, like, the new rebuild of GoChat, we paired up someone from EP with the people who had been owning, um, GoLLM.

You know, one of the things that's interesting about experimental projects is the goal is to learn quickly whether there's product market fit, whether that's with your internal users or your external users. But the goal is to basically get to escape velocity.

Like, have a product that launches and goes live, whether that's, like, GoChat GA-ing and replacing GoLLM, or whether that's, like, token billing, uh, serving our users or agentic commerce now being a thing. And what we found, and you know, this is...

The experimental projects team has been around, um, about a year and a half, and so these are relatively small samples we're talking about. But what we have found in that sort of anecdata is what we call embedded projects, projects where you take a couple people from a product or infrastructure team and a couple people from the experimental projects and group them together, are more likely to reach escape velocity.

Token billing is a good example, right? Like, we need the billing team to take it forward, and if the billing team was, like, core to it from the start, it's much easier for them to take it forward. Same thing, you know, if, if the ML Infra team, you know, deeply understands the new build and feels ownership over it, it'll be successful, um, in the long run.

So I don't think of it... You know, some people think of these teams as, like, labs teams off in the side, off in a corner, operating totally in isolation. We do some of tho- Like, we're doing that for agent pay, like agent-to-agent payments, 'cause that, that just, like-

Right

... needs a big rev on, like, what's the product shape, what's the technology shape? But wherever possible, we actually do it as a joint project, very deeply embedded with design partners, like with customers that wanna do it with us, but also with, uh, with other teams at Stripe.

Yeah, my typical line on just o- just closing the loop on the build versus buy thing, uh, is usually buy then build, if you think about the sequencing.

Yes.

Uh, I, I think your, yours is, yours is much more nuanced in terms of, like, how close-

Buy then build if a buy exists.

Correct.

If a buy doesn't exist, you might wanna build, but pick up your head of recorder to make sure you can't buy.

Yeah.

Yeah.

Yeah. Well, a- and also I think, like, mostly because I, I see the opposite when p- a lot of people try to do the opposite of build then buy to, to, to, like, reason things from first principles, but actually, uh, uh, the sheer amount of experimentation in the wider world means that a lot of people are being specialists in your thing, like evals, where you can just benefit from th- from their experience instead of, uh, reinventing the wheel.

So before Stripe, I was at Coursera. Have you guys ever... Like the ed tech platform?

Of course. Of course.

Okay. I was there for eight years, and I joined when we were less than 40 people, and it was a lot of absolutely brilliant folks from Andrew and Daphne's lab at Stanford who had never had a job before.

And by the way, I do not count myself in the absolutely brilliant folks from Andrew and Daphne's lab. I was on the East Coast, um, and not absolutely brilliant, but I had also never had a job before. And a bunch of us never had a job before, but, like really hardworking, determined people built a bunch of stuff- ...

homegrown that we shouldn't have, right?

Mm.

We had our own experimentation platform, we had our own analytics platform-

There's, there's learning value

... we had our own machine learning platform, we had our, uh... We learned a lot, we learned a lot. But, like, what is Coursera's core competitive advantage? It's not their experimentation platform. Um, and you know, that was 2014, so actually a bunch of that stuff-

Didn't exist

... didn't exist. But it was painful in 2018, 2019, 2020 to rip and replace, and rip and replace was definitely the right thing to do.

Mm.

The other sort of, um, thing I'll note on that is if you look at how Stripe is the skeletal system for all the AI companies, it very quickly becomes clear that when it comes to payments, billing, tax, revenue recognition, reporting, fraud protection, consumer checkout experience, pricing and monetization frameworks, they are completely buy.

Like, they're completely buying Stripe. That's all they're buying. Um, and I think there's an interesting thing there where w- I was talking to an AI company the other day who uses another provider to block bots at the time of signup, and they said, "It's actually really annoying to have multiple third parties doing my fraud protection.

Like, one doing it up funnel and the other doing it down funnel." Well, they asked to switch to Stripe. We don't yet have that particular functionality, but we, we could build it. But I think there's also... it's also important to reason about what is the third party that you can be, not for everything, but, like, more all in on so that it plays nice internally, so that you have fewer relationships, so that you have, like, preferential pricing, et cetera, et cetera.

And you know, when you think about cloud providers, like, that often happens a lot as well.

Vercel is definitely doing that.

Totally.

Trying to bundle.

Yeah.

So this is the economy section. We, we, we were saying you're, you're more interested in sort of the AI economy takes. Uh, the obvious big one is are we in a bubble?

Are we in a bubble? Okay, so it depends what you mean by a bubble.

Yeah.

But I think, you know, one question-

But very classic economist answer is, like, it depends on that-

It depends. I know, I know

... level.

Uh, there's al- there's no, like, two-armed economist. On the one hand, on the other hand. Okay. The question I got a lot a couple quarters ago, especially because all of these AI companies are private, is are they creating real value?

Is there real revenue coming in, right? 'Cause, 'cause it's, it's pretty clear to see that there are real costs. There's very big f- fundraises. There's a lot of capital that's flowing out. Um, is there dollars flowing in? And so that actually forced us to step back.

You know, one of the fun things at Stripe is just, like, it just-

You see the data

... you just see. You just see it going through.

You see, yeah.

Right? You see each successive wave of startups. You see, you know, people retaining and churning their subscriptions. You see who's buying what for how much from whom. Um, and, uh, when w- we stepped back and we said, "Okay, like, let's just look at this AI cohort," and there's lots of different ways to define it, but for simplicity, one cohort that we looked at was the 100 highest grossing AI companies on Stripe.

And you kind of need a reference point, and so we were like, "Let's compare them to the 100 highest grossing SaaS companies from five years prior." And we looked at things like how quickly do they get to a million or 10 million or 30 million in ARR, and the answer is two to three times as fast as a SaaS cohort.

Yeah.

We looked at questions like how diversified global is their customer base, and the answer is, at the end of their first year, at the end of their second year, basically whenever you look, they are twice as global. Uh, like, they're selling into twice as many countries.

55 countries.

They have m- majority of their revenue coming from outside their home market, even if their home market is the US. Um, and in some cases, you know, there's a startup in France who's in that list who's, whose, like, 95% of their revenue is outside of France, right?

They're very global. And then you start to look at things like retention, which also come, comes up a bunch, right? Like, is this, um, is this truly ARR, or is this, like, revenue popping and then falls off?

Times, times 12.

Exactly. And this one was a little more nuanced. So if you squint at the data, you can actually see that these AI companies, on a per-company basis, have slightly lower retention than the SaaS companies. Not, like, dramatically lower, but slightly lower, and that's also consistent with being, like, relatively early in the adoption curve, but even correcting for that, slightly lower.

But then if you bundle that, if you look at, okay, like, these SaaS companies are doing this wave of things, these AI companies are doing this wave of things, what's interesting about SaaS is the churn is churn from the entire vertical.

In AI, they're just churning from that company and flipping to another company, and then if you keep watching them, a few months later they flip back to the first company. So that tells me it's actually, like, a very competitive market.

People like the product, they wanna use the product, but there's a bunch of good products, and the best product is changing over time, and so people are, are happily flipping across. Oh, well, so in SaaS, you had this cool trend of you started horizontal.

You started with Salesforce, and then you went vertical. Like, you have the vertical SaaS, the Toasts and the whatever else. Uh, we talked a bit about wrappers earlier. AI's done the same thing. You start horizontal, right? You're, like, the infrastructure, you're the model providers, you're the purely horizontal.

And, and then all of a sudden it's, like, all of these verticals. It's like, okay, we're in healthcare and there's Nobla, and there's Ascribe, or we're in architecture and there's Studio, or we're in law and there's Harvey. Just, like, all of these verticals popping up, and popping up much faster than in SaaS.

And I think there's two things happening there. One is you can get to those verticals very efficiently because you're sitting on top of someone else's LLM, so you don't actually have to do the, the research, and it's, like, a quite lightweight build.

But the other thing is because these AI solutions are so borderless, niche markets, vertical markets at a global scale are actually quite large markets. And so there are Incentives to specialization in a way that maybe there weren't, um, five years ago.

So anyway, is it a bubble? Um- I don't know. It depends how you define a bubble. I'm a two-armed economist. But what I will tell you is, um, these are companies that are growing very quickly, faster than anything we've ever seen, very diversified in their customer base, which makes me feel better about them.

Very sticky in their customer base, not always on a per-company level, but on, like, a problem-to-be-solved level, which tells me that the customers are getting recurring value from the product and are willing to pay for it.

Yeah. What, what I'm hearing is, like, there is some real, like, better quality businesses being built. At the same time, that it has no in... Uh, the, the expectations can race ahead of those.

Yeah.

And that's not within the Stripe observable universe.

Yeah. The part we didn't... I mean, the part we didn't talk about was the cost profiles. And, you know-

Oh, the margins.

When I reason about cost profiles, there's really, like, two p- there's, like, the fixed cost and the marginal cost, or there's, like, the, the people costs and, like, the... I mean, in the case of AI, like, the inference costs, right?

And so the people cost for these AI companies are quite small, right? You look at a, a Lovable, you're talking crazy revenue milestones with 10, 20, 30, 40 people in the early days. And, and even today, right, when you, when you look at most of the top 100 AI companies on Stripe, their revenue per employee is unlike any other business, including public companies who are known for being incredibly efficient companies.

That, of course, people cost ignores the inference cost. And so I think we absolutely need to model assumptions around the efficiency and where the inference cost is going in order to be able to reason. But in the same way we were talking about how do we think about the ROI on, on these coding agents, I think we would be unwise to measure the value of these companies under the assumption of today's costs, and we need to model out based on, you know, reasonable things that we've seen and reasonable expectations we have about the world, like, those costs going down quite a bit, at which point, you know, in traditional senses, like, very interesting businesses.

Yeah. Uh, I, I would say, like, you know, there, there's the, the benign element, and then there's the less benign on, on... in terms of the cost profile, which is, yes, as, as AI is increasingly doing, uh, more and more labor that you, you would otherwise have hired for, then it should rise as, as a part of your spend.

And then there's the less benign one, which is, like, people are selling dollars for 50 cents, and that's why you're... so you're seeing such revenue traction because obviously you're kind of giving money away.

For sure. And, you know, like, I, I was a grad student in the early days of Uber and DoorDash or, you know, I was year one, two of Coursera, which was, like... That's still a .org at that time, right?

Like, basically a nonprofit. And I remember my lifestyle was subsidized by the VCs-

Oh, really?

... who were paying for part of my Uber and paying for part of my DoorDash. So, you know, we lived that pain.

Those two worked out.

Some of those prices going up. But increasingly what we're seeing from the AI companies on Stripe is they do wanna have healthy unit economics. I mean, let's not talk about, like, the big labs that are pouring crazy money into research.

But if you're talking about, like, the vertical kind of wrappers, which are themselves also, um, doing very well as businesses, they are building quite healthy unit economics. And the, the demand we've seen for token billing, I think, is actually in part a testament to the fact that they really do wanna have unit economics.

So not their overall book, but literally, like, the marginal person I serve, the revenue I get from them versus the cost I incur for them, they want those to already be in the green. So I think there are some, some very good businesses that are being built.

We've kept you for a long time. You've, you've, uh, you've indulged us in so many different topics. Uh, do you have, like, any other, like, uh, hot takes on just, like, AI and the economy that you wanna, uh, indulge in?

Like, uh, my classic hot take is, uh, how come A- AI doesn't show up in the GDP per capita numbers?

Yeah.

Right? Which is the... which is part of the whole productivity discussion. But it's really, really driving home, like, we have to see this show up somehow.

Yeah.

Right? And, and that's part of the bubble discussion. But I think to me, like, any story where technology, you know, as, as a, as a factor in the macroeconomy equation is supposed to be a big driver, you should see it in the GDP at some point.

The GDP doesn't measure everything, but, like, at some point.

We should see it in GDP. There's a lot of noise in GDP-

Yeah

... 'cause there's a lot of other drivers-

Yeah

... of GDP. How quickly we see it is-

Yeah

... I think an open question.

It's not the same year.

Yeah.

But it should be fast. Like, you know-

It should be fast-ish

... these things, these things... A- and, like, the, the, the funny hot take is, like, the only re- the only way we're seeing GDP is the data center build-outs. That's-

Oh, interesting. Oh, yeah. I'm less close to that. It could be. I mean, hot takes. I think AI should make markets more efficient. I think agents should make commerce more efficient, which should genuinely expand the aperture of what people buy.

I think agents are already making business creation more efficient, which should accelerate new startup growth, which we are also seeing. And, you know, if, if you just think about, like, the tens of thousands, hundreds of thousands of businesses that are getting started in these AI dev tools that, like, would not have been getting st- started before, I think that's incredibly promising.

I think we have a real cost question on our hands, but I, I think it will be solved, um, in... for many domains. I think there will be cheap enough models to do the job that create meaningful value.

I think the AI companies are being quite savvy. We talked a bit about the unit economics, but also, like, what are they pricing to, right? So SaaS was mostly seat-based, and you could imagine sort of a death spiral where, like, AI is seat-based, but AI is replacing the seats, and so you need fewer seats- ...

and you monetize less. Like, you don't wanna peg your revenue to the thing you're trying to replace, right? Um, and I think, you know, outcome-based or usage-based will be, will be much more powerful. I'm seeing more adoption of AI Outside of the US in a broad range of markets than I expected.

Um, some of that-

Brazil, Brazil is huge. I'm about to go over-

Brazil is huge

... yeah.

It's a little bit hard to parse what is, like, well, Stripe is opening up more markets and having more payment methods and giving them more exposure versus, like, literally there is an expansion in adoption. But I think it will be promising for the world if there's more...

It's not really equality of access 'cause, like, on paper anyone has access, but, like, equality of adoption so that, uh, we don't lead to sort of- we don't end up with very uneven economic growth as a result of AI.

So we'll, we'll have to watch. I don't think it's gonna show up next year. I don't think most businesses are targeting employee efficiencies next year, but I think every business is targeting employee efficiencies for '27 and for '28-

Yeah

... which is suggesting more efficiency. And if you can couple more efficient production with more efficient consumption, which frankly is what agents do, then one would expect, yes, GDP to rise, to rise meaningfully. And, and I think you could debate is it one percentage point more growth per year?

Is it three percentage points more growth per year? Like- ... I don't know. I don't think it's 10. I don't... Well, I mean, I'd love it, but I don't think so. Um-

I think we'll-

I think, I think we'll have to see.

People-

'Cause that thing compounds, right? Like-

That thing compounds

... that thing compounds.

GDP's a big number.

Yeah.

Uh, but yeah, no, the, the term I've had for the, for the, the movement of employee efficiency is, uh, tiny teams.

Mm-hmm.

Like, uh, you know, teams with more millions in revenue than employees, uh, which, which, like, completely changes the startup structure because you are profi- probably profitable from, like, maybe your first round of funding.

Yeah.

That changes everything.

I have one more hot take, which is it's easy to think in a world of, like, really exciting, powerful tech that somehow brand doesn't matter, it's all about the technology. But actually-

Yeah

... if you look over-

Brand matters a lot

... the last year-

Yeah

... so much of the, you know, value created in AI companies has actually come fr- And Lovable was brilliant in branding themselves Lovable, right?

Yeah.

Like, so many of these wrappers are actually winning on a really differentiated user experience, and a really compelling brand, and a really compelling community. And so I don't know. I just, I just, I, you know, sometimes, you know, investors, friends, whatever, will be like, "Oh," like, "what do you think of this?

What do you think of that?" And it's like, you know, great founders who are highly technical are amazing, but you also need them to be hyper-focused on, like, the user and the product experience and, um, really creating something, like, beautiful and crafty and, and I think some people are like, "Oh," like, "AI is gonna replace that," and, like, "All that matters is the tech, and there's not gonna be a human internet, and it's gonna be agents talking to agents."

And it's like, I don't know, maybe, but, like, so far what I'm seeing is brand matters more than ever.

Yeah. The, the, the Silicon Valley phrase is you need rizz and tizz, and if you don't have rizz, then -

Exactly

... then, then that's all. And I think, like, that Stripe, something Stripe has always embodied, like very good technologies with also, uh, industry-leading design, uh, which I think is very important.

I'm a Katie Dill fangirl.

Uh-

Katie Dill's our head of design.

Oh, okay.

Actually, I, I was co-hosting Friday Fireside, which is, like, our weekly-

Ah

... company thing today, and, um, I, I cited something that Katie's team did and made it clear that I was a fangirl. And Tanya, who's our head of PMM, um, said she was gonna have to fight me for, like, uh-

Head, head girl

... Katie's biggest-

Yeah.

... Katie's biggest fan. Anyway, we decided that the three of us would just go to a spa and- ... drink mimosas-

Ah

... instead of fighting. But for a hot second in the- ... in the, in the company Slack chat there was, like, a, maybe a fight between TK and I.

Okay. Let me... While you're on this topic- ... what's, what's something that you learned from her that, like, has really driven design at Stripe?

Yeah. Katie does not give an inch on quality.

Okay.

And it doesn't matter if it's, like, one banner that 2,000 users see that, you know, has some font that's slightly bigger than it should be. Like, that's a bug.

Yeah.

That has an SLA. That needs to be burned down. And actually, now every two weeks we have a run the business review where it's, like, us 60 folks get together and, and talk about the whole business. And, and literally each of us has a slide that's like, "Did we meet our bug burn down SLA for these?"

Like, largely qual- not exclusively quality, but often quality issues. And I think, um, she builds, like, beautiful exper- she and her team design, like, beautiful experiences, and sort of like macro are, like, extremely innovative. But there's something that I've learned around, like, the micro.

Like, you have to obsess over every detail, and one tiny thing that's not good enough is worth all of us sweating until it is good enough. It can be a little exhausting at the scale of Stripe, but it's also, like, very grounding to just know, like, there's a clear line, and if it doesn't meet the quality bar, like, you just gotta fix it.

Wow. Well, thank you for, uh, spending some time with us and explaining how things work at Stripe. I mean, like, we... Everyone's always curious, and you've been very generous with your time.

Oh, thanks for having me, and, uh, it's been really fun.

Uh, call, call to action. Uh, hiring, I assume?

Yes. I, we are hiring. We are hiring. I mean, we're hiring everybody.

Yeah.

Uh, but we are, but we are particularly hiring, um, machine learning engineers/scientists. A lot of back-end folks, so, like, if you're excited to, you know, build the infrastructure for building agents or build the infrastructure to do machine learning, a lot of those.

Um, we are recognizing that data is, like, increasingly an asset, that our users want real-time, and high quality, and well-documented. So, like, if you're big on data engineering or building data platforms, um, also hiring there. But just, like, across the stack.

It's, like, a great team and fun projects, so yeah, we're hiring.

Excellent.

Cool. Thanks, Emily.

Awesome. Thanks, folks.