Designing REST APIs for the age of AI agents
summary
AI agents start driving API traffic and require more than endpoint names: the talk shows that OpenAPI 3.5 is necessary but not sufficient, recommends richer request/response schemas, descriptive error payloads and idempotency support, and demonstrates a working web tool prototype (Chrome + tool) that orders coffee. It argues for adaptive rate limiting over fixed quotas, recommends publishing specs (including agent.json / web MCP) for discoverability, and reports successful agent-driven ordering with 429 and retry semantics handled.
AI-written summary · about
chapters (10)
- 0:00REST APIs and AI agents
- 4:48AI agents as API consumers
- 8:45Machine-readable API specs
- 14:43Consistency in API design
- 17:35AI agents navigating APIs
- 22:35Efficient API resource structuring
- 25:27Idempotency and rate limiting
- 32:54Every company is an API
- 37:51APIs for AI agents
- 40:43APIs for humans and AI
related talks
23mDesigning APIs Both Humans and AI Love
24mThe API Economy's Next Chapter: AI Agents as First-Class Clients
13mAre Your APIs Ready for AI Agents?
35mUnleash Your Inner API Architect: Build Agents That Deliver
26mIf your APIs aren’t agent‑ready, your AI agent is just chat: [...]
24mAPIs Are the Nervous System of Agentic AI - Designing Agent-Ready APIs for Aut…
21mUsing the Power of APIs and Domain Driven Design to Embrace AI [...]
45mFishbowl Discussion on APIs and AI