Transcript: ‘How Stripe Is Building for an Agent-native World’
Summary
Stripe's Head of Data and AI, Emily Glassberg Sands, discusses the profound shift in the internet economy from human-centric transactions to an "agent-driven" model where AI agents increasingly act as buyers, sellers, and builders. This transition introduces new challenges, particularly in fraud detection, where "compute theft" has emerged as a significant risk, with multi-account abuse and free trial exploitation costing AI companies substantial amounts. Stripe has expanded its Radar fraud detection from checkout to the full customer lifecycle, integrating at signup and throughout customer interactions. Sands highlights that AI companies are scaling revenue significantly faster than previous SaaS cohorts, reaching $30 million ARR in 18 months, three times faster than top SaaS companies from 2018. This rapid growth is fueled by net-new spending and a hyper-experimentation with monetization models, moving from seat-based to usage-based and outcome-based billing, exemplified by Stripe's "token billing" solution. The discussion also covers the evolution of developer experience to accommodate agents as builders and the spectrum of "agentic commerce," from AI-assisted buying to autonomous purchasing, facilitated by protocols like the one co-created with OpenAI and Stripe's consumer wallet, Link, which is being adapted for delegated agent purchases with guardrails.
Key takeaway
For CTOs and VPs of Engineering building AI-first products, recognize that traditional fraud detection and billing models are insufficient. Your teams should proactively integrate full-lifecycle fraud protection, such as Stripe Radar at signup and for overages, to mitigate compute theft and free trial abuse. Additionally, re-evaluate your monetization strategy, moving towards usage-based or outcome-based billing to align with AI's marginal costs and customer value, as seat-based models will likely become obsolete in the enterprise.
Key insights
The internet economy is rapidly shifting to an agent-driven model, necessitating new approaches to fraud, billing, and commerce infrastructure.
Principles
- Compute theft is the new payment fraud.
- AI companies scale revenue faster than SaaS.
- Outcome-based billing will replace seat-based pricing.
Method
Stripe expanded its Radar fraud detection to cover the full customer lifecycle, from signup through overages, by integrating its API at multiple interaction points beyond just checkout.
In practice
- Integrate Stripe Radar at signup for AI services.
- Consider outcome-based pricing for vertical AI solutions.
- Utilize shared payment tokens for secure agent transactions.
Topics
- Agent-Native Economy
- Compute Theft Fraud
- Stripe Radar
- AI Company Growth
- Outcome-Based Billing
Best for: Investor, CTO, VP of Engineering/Data, Director of AI/ML, AI Product Manager, Entrepreneur
Related on AIssential
Editorial summary, takeaway, and curation by AIssential. Original article published by AI & I - Every.