"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis Stripe's Payments Foundation Model: How Data & Infra Create Compounding Advantage, w/ Emily Sands
60 snips
Sep 25, 2025 Emily Sands, Head of Data and AI at Stripe, shares her insights on building a payments foundation model that processes vast transaction data for enhanced fraud detection. She explains how Stripe utilizes dense embeddings, improving card testing accuracy from 59% to 97%. Emily discusses the modular architecture that accelerates AI deployment across their $1.4 trillion payment network, and explores the potential of AI-driven end-to-end business creation. Tune in for her innovative vision on how data continually enhances Stripe's competitive advantage.
AI Snips
Chapters
Transcript
Episode notes
Payments As A Native Modality
- Stripe trained a transformer to turn every payment or short sequence into a dense vector embedding for reuse across tasks.
- Those embeddings let teams build lightweight classifiers and explanations without retraining large models.
Structure Enables Dense Learning
- Payments data is highly structured so Stripe built a custom tokenizer to compress numeric and categorical signals efficiently.
- Modeling short histories across cards, devices, merchants yields far richer signal than single-transaction features.
Dense Network Is A Differentiator
- Stripe's network is extremely dense so many cards and entities reappear across merchants, reducing hops needed for detection.
- Dense cross-merchant visibility yields disproportionate fraud and conversion signal.