Anthropic’s Claude agent lead says delegation, not prompting, is now the bottleneck
petergyang · x · 2026-07-26
Jess Yan, product lead for Claude Managed Agents at Anthropic, argues that agents have moved beyond prompt loops into autonomous, long-running coworkers that can hold permissions, access internal tools and sensitive data, and decide when to keep going versus ask a human.
Core thesis
- The bottleneck is no longer just individual productivity; it is how much work you can delegate safely at once.
- The highest-leverage path is to make one person exponentially more powerful, then expand from there.
What changed in agents
- They are now self-recovering, persistent actors rather than simple Q&A loops.
- A successful agent stack needs a harness: memory, permissions, internal tool access, and escalation logic.
- The product challenge is no longer just prompting, but building reliable scaffolding around long-running work.
More from coding & agent
- Shopify design lead on AI-era workflow: designers are becoming full-stack editors — KlausCodes · 2026-10-26
- Ultrametric open-sources Open Startup Repo to let AI agents run startup operations — garrytan · 2026-10-26
- MCP fault injection study: agent task scores drop 60% under realistic tool failures, scale doesn't help — dair_ai · 2026-10-26
- General agents burn ~140M tokens/month on LinkedIn outreach; scripts do it 60x cheaper — ChrisHarpon2 · 2026-10-26
- OpenAI upgrades Decisions API: 400 questions per call, native voice and image support — OpenAIDevs · 2026-10-26
- NLI alone rejects 5/6 new agent memories; a 0.45 similarity gate fixes it all — Alternative_Pin9598 · 2026-10-26