Jul 10, 2025 · 1h 15m · mad
The Rise of Agentic Commerce — Emily Glassberg Sands (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
In this episode of The MAD Podcast, host Matt Turck interviews Emily Glassberg Sands, Head of Information at Stripe, discussing Stripe's proprietary AI foundation model, internal AI culture, and the rapid rise of agentic commerce. They analyze macro data from Stripe's platform showing how modern AI startups achieve unprecedented monetization speed, global expansion, and vertical specialization.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Matt holds 15.2% of the talking time here. How this is scored →
Matt as informed peer 3.7 Guest teaching 5.1 Guest disagreement 0.5 Matt pushing back 0.4
speaking balance: gold is Matt, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 20:18 Reframing the naive model approach
Emily directly rejects the initial approach Stripe took of scaling wider transformers on isolated payment tokens, explaining that their first instinct was completely wrong before pivoting to sequence modelling.
Hardest push from Matt ▶ 16:24 Challenging foundation model replacement of MLMatt presses Emily on whether foundation models are outright replacing traditional machine learning models or if she is jumping to conclusions about ensemble approaches.
Biggest teaching moment ▶ 49:51 Structural transformation of agentic commerceEmily educates Matt on how AI agent purchasing behavior differs fundamentally from human browsing, requiring structured intent endpoints, machine-readable schemas, and scoped credentials.
Matt holds his own ▶ 13:16 Probing payment data heterogeneityMatt demonstrates deep technical domain understanding by highlighting the structural differences between natural language and sparse, non-grammatical credit card transaction data.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Matt as informed peer | Guest teaching | Guest disagreement | Matt pushing back | Why |
|---|---|---|---|---|---|---|
| Stripe's Operational Scale and Emily's Remit | 2 | 3 | 1 | 0 | Matt opens with broad questions regarding Stripe's business scale and Emily's background. Emily outlines key operational metrics like processing 1.3% of global GDP and details her information group remit. | |
| Building Stripe's Proprietary Payments AI Foundation Model | 5 | 6 | 1 | 2 | Matt presses on why Stripe built a proprietary foundation model instead of using general LLMs and probes model architecture. Emily explains using BERT encoders over GPT decoders and how unsupervised payments embeddings boosted card testing recall from 59% to 97%. | |
| Explainability, Risk Rules, and Smart Disputes | 5 | 6 | 1 | 1 | Matt inquires about explainability, regulatory transparency, and internal data infrastructure choices. Emily outlines dynamic risk rules, Smart Disputes automated evidence gathering, and lessons learned migrating ML infrastructure to Shepard. | |
| The Rise and Architecture of Agentic Commerce | 4 | 7 | 1 | 1 | Matt asks how Stripe handles autonomous shopping agents and multi-agent coordination. Emily explains the shift from human to AI agents using virtual cards and details future architectural needs like intent endpoints and machine-readable product schemas. | |
| Model Context Protocol (MCP) and Stripe's Implementation | 4 | 5 | 0 | 0 | Matt explores where Model Context Protocol (MCP) fits into agentic architecture. Emily shares how Decagon built an integration in under a week to handle support automation, reducing support costs by 65%. | |
| Rapid Revenue Growth and Monetization Trends of AI Startups | 3 | 6 | 0 | 0 | Matt asks about macro trends across AI startups on Stripe. Emily highlights that top AI startups reach $30M ARR in 1.5 years compared to 5.5 years for historical SaaS benchmarks. | |
| Global Footprint and Lean Scaling in Early-Stage AI Companies | 4 | 5 | 0 | 0 | Matt asks whether AI startups go global earlier than previous tech waves. Emily reveals median AI startups sell into 55 countries in year one, while Matt highlights how abstracted infrastructure like Stripe and AWS enables lean global growth. | |
| How Global Scale Drives AI Niche Specialization and Verticalization | 4 | 6 | 1 | 0 | Emily presents a hypothesis that borderless operations make vertical specialization lucrative. She also details how AI startups are shifting from per-seat pricing to usage-based and outcome-based billing. | |
| Internal AI Adoption, Tooling, and Literacy Culture at Stripe | 4 | 5 | 0 | 0 | Matt brings up recent executive memos on AI literacy and internal adoption. Emily describes Stripe's internal bottoms-up LLM Explorer deployment and prompt preset sharing ecosystem. | |
| Stripe's Product Roadmap and Future Vision for AI Commerce | 2 | 2 | 0 | 0 | Matt prompts Emily for roadmap teasers to wrap up the interview. Emily summarizes upcoming focus areas around payments foundation models, risk as a service, and agentic commerce. |
Statements from this episode (26)
Stripe processed $1.4 trillion in 2024, representing 1.3% of global GDP
“Last year, companies processed about 1.4 trillion dollars on Stripe. To put that in perspective, because it's a lot of zeros, that's about 1.3% of Global GDP. And that number grew 38% year over year in what many experienced as kind of a rocky macro climate.”
Stripe handles an average of 50,000 new transactions every minute
“Stripe's network handles on average about 50,000 new transactions every minute.”
Businesses using Stripe grew seven times faster than the S&P 500
“Businesses on Stripe grew seven times faster last year than the S and P 500.”
Stripe's shared embeddings reduce model development time from quarters to weekends
“Makes spinning up a new model become A weekend project, not a quarter project, because you get kind of out of the box these embeddings.”
AI foundation models will eventually fully replace traditional machine learning models
“I think we will get to a point where it fully replaces.”
Stripe's AI foundation model increased card testing fraud detection to 97%
“Our detection rate on large merchants went from 59% to 97%.”
Merchants lose approximately $55 billion annually to chargebacks
“Merchants lose about fifty five billion dollars a year to chargebacks.”
Vimeo and Squarespace recover 13% more disputed revenue using Stripe AI
“And like Vimeo and Squarespace were our two first adopters, but they're recovering 13% more revenue on disputed charges from adopting it, and they're doing that with zero extra labor.”
Turo recaptured over $100 million annually by adopting Stripe Checkout
“So like Turo, maybe you've used it there like a, the world's largest car sharing marketplace. They moved over to our checkout suite and saw a five percent increase in recaptured revenue, which for them was, I think like a hundred and some million dollars a yea…”
Displaying non-card payment options increases merchant revenue by roughly 12%
“Like businesses that show at least one relevant payment method beyond just cards see like a 12% increase in revenue and more than seven percent lifting conversion.”
Applied AI startups often lack the security controls enterprise buyers require
“So there are often new startups, less so on the infrastructure side and more so on the applied side, who we would love to buy from, partner with, but they don't have the security protocols and controls in place for us to feel comfortable operating in their sta…”
Perplexity launched in-app hotel discovery and booking powered by Stripe
“So like perplexity, you may have seen that they recently launched hotel discovery and booking in the app and it's powered by Stripe”
Merchant APIs will require canonical intent endpoints for autonomous AI agents
“And so every merchant API is probably going to need one canonical kind of intent endpoint that accepts those structured desires instead of sort of this UI click world that we live in today.”
Merchants must expose machine-readable product schemas to sell through AI agents
“And so I think early adopters who want to sell through agentic channels are going to need to expose kind of an open product schema, like the SKU and the inventory and the price and the constraints and, you know, maybe even the wedge that you're willing to give…”
AI agents will shrink e-commerce latency budgets to a few hundred milliseconds
“I think latency budgets are gonna shrink to machine time. We talked about latency budgets in the context of the charge path, but, like, you know, people will wait three seconds for a spinner. I think an agent's just gonna retry somewhere else after a couple hu…”
Decagon's first Stripe MCP integration customer saw support costs drop 65%
“And the first Decagon customer that they released this to reported a 65% drop in support costs.”
Model Context Protocol is becoming the default standard for LLM integration
“I mean, it's pretty clear that MCP is becoming the default way that any single service Stripe or GitHub or Notion talks to an LLM.”
78% of the Forbes AI 50 list use Stripe for payments
“We recently looked at the Forbes AI-Fifty and 78% of them are Stripe users. That 78% reflects 100% of the Forbes AI-Fifty that accept online payments.”
AI startups reach $30M ARR three times faster than fast-growing SaaS startups
“Those that already hit thirty million in annualized revenue got there in about a year and a half. For comparison, you know, many of us were around five years ago, like the fastest growing SaaS startups on Stripe took, you know, five and a half years to hit tha…”
Stockholm-based AI startup Lovable reached $50 million ARR in six months
“European breakouts lovable out of Stockholm hit fifty million ARR in six months and is now for sure the fastest growing startup in Europe.”
AI coding assistant Cursor passed $300 million ARR two years after launch
“Cursor, which, you know, of course we mentioned earlier, helps developers code with AI. They only launched two years ago. They recently announced that they're over three hundred million in ARR.”
Median top AI startup expands to 55 countries in its first year
“AI, that AI 100 group, and you ask the median is in 55 countries in their first year, and 80 countries by their second year. And that is twice the internationalization of equally promising earlier SaaS companies at the same stage of their evolution.”
The median top AI startup generates 56% of revenue internationally
“Like, today, these companies generate the majority of their revenue. I think the median is 56% of revenues from international Customers.”
AI photo editor PhotoRoom reached $50 million ARR in three years
“They went from, I think, zero to fifty million ARR in three years. They already sell into a 184 markets.”
AI software pricing will shift to outcome-based models within five years
“I think it's where actually like the market equilibrium, like where clearing will actually happen, you know, two, three, five years from now is experimenting with new pricing models, like outcome based pricing and actually increasingly using outcome based pric…”
Intercom is shifting support software pricing from per-seat to per-resolved-case
“Intercom is an Irish founded company, and they're also reinventing customer service. There's a lot of interesting stuff in the customer service space, but they're moving their support product from charging per seat, right? The olden days model, which is how mo…”