LangChain Founder: Evals Work for Narrow Tasks but Break Down for Autonomous Agents
hwchase17 · x · 2026-10-09
LangChain founder hwchase17 responds to Hunter Gerlach's loop: 1) create high-value evals, 2) realize you're fighting Goodhart's Law, 3) repeat. His initial take: eval-driven development works for narrowly scoped tasks but fails for more autonomous agents — raising the open question of how to actually hill-climb those.
More from coding & agent
- Apple Paper: SOTA Harnesses Offer No Edge Over Minimal Single-Session Coding Agent — himanshustwts · 2026-10-09
- PartyKit shuts down free hosted platform 2.5 years after Cloudflare acquisition — threepointone · 2026-10-09
- Run a Brand X Account With 3 AI Bots: Community Manager, Content Strategist, Analyst — FinanceYF5 · 2026-10-09
- Google's A2A protocol moves to neutral foundation alongside MCP under AAIF — SnooDingos9560 · 2026-10-09
- Who owns the job after the agent disconnects? A Kubernetes MCP architecture — stevenacreman · 2026-10-09
- Deepkit author: Bun is rediscovering our decade-old ideas, agents could revive it — MarcJSchmidt · 2026-10-09