Create AI Agents That Work Independently
summary
Claims an AI agent can autonomously complete multi-step workflows—book a flight then a hotel—while keeping context across days using persistent state; uses Cloudflare Durable Objects for low-latency, long-term memory (states survive 24+ hours) and supports pausing idle agents to save resources. Identity and authorization rely on providers like Auth0 or Stytch or WorkOS so agents can securely call external APIs. Coordination follows the Model Context Protocol (MCP) to connect tools and manage authenticated access end to end.
AI-written summary · about
chapters (10)
- 0:00Introduction and welcome
- 0:47What are AI agents
- 2:11Identity memory and coordination
- 4:17Authentication challenges
- 5:20Identity providers in AI agents
- 6:45Durable objects for memory
- 7:50AI's impact on developers
- 8:54Future of MCP and AI agents
- 9:16Autonomous agents in the future
- 9:57Advice for developers
related talks
31mHow to Build Autonomous AI Agents That Run Anywhere
24mAI Agents in Production: Why the Platform Matters
34mAI Agents: Transforming Autonomous Decision-Making
29mAgentic AI: What It Is and Why It Matters
27mAgentic Engineering: Working With AI, Not Just Using It
29mWhat exactly are AI Agents?
26mAI agents 101
40mIn AI We Trust? Securing the Future, One Agent at a Time