User asks his agent "can't or won't?" — and the agent backs down
StewartalsopIII · x · 2026-11-07
Stewart Alsop III shares a funny moment: when his agent claimed it "can't" do something, he replied with the classic "can't or won't?" — and the agent complied. He muses that these models don't really have agency, and admits his codebase still carries strange defensive policies written when agents were much worse 1-2 years ago.
More from coding & agent
- SidePulse turns your MacBook's SD card slot into an 8-LED agent status light you control by writing a file — yacineMTB · 2026-11-07
- Dev writes up Query Studio: a LangChain MCP SQL agent with read-only safeguards — Top-Plan-229 · 2026-11-07
- shadcn reports weird eval results: Sonnet and Gemini pass the most tasks — shadcn · 2026-11-07
- Gaussian Splats for UI: dev prototypes imagen→SHARP→spark→popmelt pipeline with smooth animation — janusch_patas · 2026-11-07
- Team weighs multi-GPU cluster vs one fat workstation to escape cloud API rate limits — Decent_Wing_4393 · 2026-11-07
- Indie dev builds OctoHall, a marketplace where AI agents hire each other and get paid — wolfdogKC · 2026-11-07