Oct 30, 2025 · 1h 37m · latent-space
The Agents Economy Backbone - with Emily Glassberg Sands, Head of Data & AI at Stripe
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Emily Glassberg Sands, Head of Data and AI at Stripe, discusses the foundational infrastructure powering the emerging AI and agentic economy, spanning open commerce protocols, real-time foundation models, and dynamic monetization architectures. She details how Stripe combats compute abuse, accelerates internal engineering velocity, and scales real-time data platforms for an autonomous future.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The hosts hold 3.8% of the talking time here. How this is scored →
The hosts as informed peer 5.1 Guest teaching 5.8 Guest disagreement 1.2 The hosts pushing back 1.9
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 57:22 Emily demands mandatory LLM citations and rejects slop
Emily firmly states a non-negotiable policy requiring staff to disclose LLM generation, voicing intense frustration with deceptive, unrefined text output.
Hardest push from the hosts ▶ 59:06 Swyx defends AI drafts and slopSwyx openly dissents from Emily's strict anti-slop stance, arguing that humans produce slop as well and that what matters is the human editor taking final accountability.
Biggest teaching moment ▶ 1:21:40 Emily reveals Stripe's AI cohort velocity dataEmily educates the hosts with proprietary Stripe data showing top AI startups scaling to ARR milestones 2 to 3 times faster and reaching twice as many global markets as prior SaaS cohorts.
The host holds their own ▶ 40:30 Swyx contrasts ACP with agent wallet architecturesSwyx demonstrates deep domain knowledge of competing protocol designs by contrasting Stripe's payment tokens against Solana and Circle's autonomous agent wallet infrastructure.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The hosts as informed peer | Guest teaching | Guest disagreement | The hosts pushing back | Why |
|---|---|---|---|---|---|---|
| Evolution of Foundation Models and Card Testing at Stripe | 4 | 5 | 1 | 1 | Alessio and Swyx ask about Stripe's transition from traditional ML to domain-specific foundation models. Emily explains how Stripe runs foundation model inference across 50,000 transactions per minute under 100ms latency to detect card testing hidden in large merchant volumes. | |
| Combating Compute Abuse and Friendly Fraud in AI | 5 | 6 | 1 | 1 | Alessio asks about the economic scale shift of fraud between SaaS and AI compute costs. Emily explains friendly fraud vectors like disposable free trial cards and large refund abuse that threaten AI startups with real marginal compute expenses. | |
| AI Monetization Architectures: Token Billing to Stablecoins | 5 | 6 | 1 | 2 | Swyx probes into monetization mechanisms and economic concentration indices. Emily details Stripe's real-time token billing API, outcome-based billing with Intercom, and high-value cross-border stablecoin settlements. | |
| Introducing the Agentic Commerce Protocol with OpenAI | 4 | 6 | 1 | 1 | Swyx transitions into the Agentic Commerce Protocol (ACP) announcement with OpenAI. Emily details why an open standard with shared payment tokens was necessary so merchants could expose inventory without agents absorbing liability. | |
| Managing High-Demand Drops, Pre-Checkout Bot Detection, and Market Efficiency | 6 | 5 | 1 | 3 | Alessio and Swyx explore bot mitigation during scarce drops and suggest auction mechanisms for market clearing. Emily explains pre-checkout fraud signals and cart-locking constraints that standard auctions do not resolve. | |
| Protocol Architecture, Agent Wallets, and Consumer Time Frontiers | 6 | 6 | 1 | 2 | Swyx questions why Stripe chose an open protocol rather than a proprietary agent wallet like crypto alternatives. Emily discusses Dwarkesh's insight on consumer time constraints and contrasts virtual card issuance against payment tokens. | |
| Accelerating Developer Monetization Through Claimable Sandboxes | 4 | 5 | 1 | 1 | Swyx asks about enabling agents to receive revenue. Emily outlines claimable sandboxes integrated directly into AI dev tools like Vercel and Replit to allow instant business initialization. | |
| Internal AI Adoption and Rapid Payment Method Integrations | 4 | 6 | 1 | 1 | Alessio inquires about internal AI workflows at Stripe. Emily details how LLM tooling compressed local payment method integrations from two months down to two weeks, while monitoring engineering assistant costs. | |
| Workplace Culture, Rigorous Thinking, and Resisting AI Slop | 6 | 5 | 4 | 5 | Alessio and Swyx challenge memo-writing culture and defend AI drafts. Emily adamantly argues that writing is deep thinking and requires LLM citations to prevent low-effort slop, prompting Swyx to explicitly push back as a pro-slop editor. | |
| Internal Tool Shed MCP, RAG Integration, and Multi-Year AI ROI | 5 | 6 | 1 | 1 | Swyx asks whether RAG remains relevant alongside internal tools. Emily describes Stripe's internal Tool Shed MCP server and explains why AI tooling ROI must be amortized over two to three years rather than in-year. | |
| Data Infrastructure: Hubert Text-to-SQL and Semantic Events | 6 | 6 | 1 | 1 | Alessio and Swyx discuss text-to-SQL accuracy and semantic layer architectures. Emily reviews evaluation metrics for their internal assistant Hubert, explaining why data discovery on uncurated schemas is the hardest hurdle. | |
| Stripe's Build vs. Buy Strategy and the Spotlight Program | 5 | 6 | 1 | 2 | Swyx introduces his 'buy then build' framework. Emily outlines Stripe's Spotlight RFP program, its joint Cronon development with Airbnb, and why embedded cross-functional teams prevent sunk-cost traps. | |
| AI Economy Analysis: Growth, Churn Dynamics, and Unit Economics | 6 | 7 | 1 | 3 | Swyx asks if the AI ecosystem is a bubble with subsidized pricing. Emily shares data from Stripe's top 100 AI cohort showing ARR growth 2-3x faster than SaaS, higher global reach, and churn representing vertical hopping rather than category abandonment. | |
| Macro AI Projections, Brand Importance, and Stripe Hiring Call | 5 | 6 | 1 | 2 | Swyx questions why AI productivity is missing from GDP metrics. Emily explains macro timeline lags, cautions against seat-based monetization, and emphasizes why brand and micro-craftsmanship remain essential. |
Statements from this episode (35)
Sands: Stripe processes $1.4 trillion annually, roughly 1.3% of global GDP
“And when you think about what we're looking at on the order of 1.3% of global GDP. About 1.4 trillion dollars a year is processed on Stripe.”
Sands: Stripe processes approximately 50,000 transactions per minute
“I mean, we are, we see like 50,000 transactions a minute.”
Sands: Stripe runs domain foundation models with sub-100ms latency on charges
“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 you can block them. So yes, it is happening on the charge path…”
Sands: Roughly 47% of payments leaders cite friendly fraud as biggest issue
“Actually, 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.”
Sands: High inference costs make friendly fraud existentially threatening for AI startups
“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 rack up these charges and you never pay, is like existentially threatening for AI businesses.”
Sands: All Forbes AI 50 Companies Monetizing Online Use Stripe
“So if you look at the Forbes AI-fifty, all of the Forbes AI-fifty who monetize online monetize through Stripe.”
Sands: Top AI Startups Expand Internationally Twice as Fast as SaaS
“Like we were looking at the top hundred grossing AI companies on Stripe and like the median was in 55 countries at the end of their first year and over a hundred countries at the end of their second year, which is like Twice as global as the SaaS wave from thr…”
Sands: Stripe Link Surpasses 200 Million Consumers
“Like, Link just passed, two hundred 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.”
Sands: 58% of Lovable's Payment Volume Flows Through Stripe Link
“Lovable 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 …”
Sands: Over 1M Shopify merchants and Salesforce joining ChatGPT checkout
“There's over one million Shopify merchants coming soon, including some really big ones like Glossier and Viore.
this week, Salesforce announced that they're also in”
Sands: Walmart and Sam's Club signed up for ChatGPT Agentic Commerce Protocol
“In the last couple of days, Walmart and Sands Club have just signed up to also make their inventory purchasable
Through ChatGPT 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 W…”
Sands: AI Agents Will Make Fraud Decisions Within Six Months
“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 mo…”
Sands: Stripe Merchants Grew Seven Times Faster Than S&P 500
“Last year the companies on Stripe grew seven times faster Than the S and P 500.”
Sands: Agentic commerce will expand total consumption by removing time constraints
“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 for spending by high-income people because it's removing the most costly o…”
Sands: Perplexity's travel search and booking agent is powered by Stripe
“A year ago, Perplexity launched a travel search and booking agent. Did you guys see this? Yep. That is also powered by Stripe.”
Sands: AI microtransactions will force agent payments beyond traditional card rails
“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 inference or buying content. You want to be able to make five, 10, 25, fifty-cent transactions, and …”
Sands: 8,500 of Stripe's 10,000 employees use LLM tools daily
“8500 stripes a day use. Are LLM-based tools. Okay, there's like only 10,000 Stripes.”
Sands: LLMs will cut Stripe payment integrations to two days
“The LPM team, it took them two weeks for the first one, but they just launched a new pan-European payment method, which 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, wi…”
Sands: 65% to 70% of Stripe engineers use AI coding assistants daily
“65, 70% of engineers use them on the day to day.”
Sands: Manual writing forces first-principles reasoning that LLMs dangerously bypass
“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 cons…”
Sands: Citing LLM usage in workplace documents is a non-negotiable rule
“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.”
Sands: Stripe built central MCP server 'Tool Shed' for internal AI tooling
“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 to like all the Stripe tools. And, you know, what I like about that is like, and it's managed centrally, li…”
Sands: Switching to o3-mini saved Stripe $3M yearly on risk task
“Or we use, like GPT-IV-O, 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, O-three mini is out, and it's, like, you know, three million dollars a…”
Sands: Enterprises should measure AI adoption on 2-3 year ROI
“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, …”
Sands: Stripe's internal text-to-SQL AI agent Hubert has 900 weekly users
“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 ar…”
Sands: Static dashboards will be obsolete within nine months as agents take over
“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 going to want to go and like look at a even like static dashboard and click around. They're going to want…”
Sands: Stripe chose Braintrust for AI evaluations over two dozen vendors
“So we actually had like more than two dozen applicants for this evals RFP. Shawn 'Swyx' Wang: There's no way you can evaluate all of them. Emily Glassberg Sands: Well, we actually, we did. So they wrote like nice one pagers. We read them all. We narrowed it do…”
Sands: Stripe passed on Tecton, co-building Cronon with Airbnb for latency
“We evaluated Tekton. This was a couple of 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 got to be operating at six nines. We got to be like, Decisioni…”
Sands: Innovation teams reach escape velocity when co-staffed with existing product engineers
“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.”
Sands: Top 100 AI startups reach ARR milestones 2-3x faster than SaaS
“One cohort that we looked at was the hundred 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 hundred highest grossing SaaS companies from five years prior. And we looked at things …”
Sands: Top AI companies have slightly lower per-company retention than SaaS
“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.”
Sands: AI churn reflects switching between competitors rather than abandoning category
“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.”
Sands: Top 100 AI startups on Stripe have unprecedented revenue per employee
“When you look at most of the top hundred AI companies on Stripe, their revenue per employee is Unlike any other business, including public companies who are known for being incredibly efficient companies.”
Sands: Vertical AI wrapper companies are building healthy unit economics
“Increasingly what we're seeing from the AI companies on Stripe is they do want to have healthy Unidec. 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, wh…”
Sands: Most businesses target AI employee efficiencies for 2027 and 2028
“I don't think it's going to 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, which is suggesting more efficiency.”