Dev vents: 90% of agent work goes into building guardrails to catch its own mistakes
alexcovo_eth · x · 2026-09-02
A developer argues that autonomous agents doing real work feels structurally backwards: most of the spend (tokens + time) goes into checks, gates and plumbing to prevent mistakes — by his estimate, 90% of the work is building guardrails against the agent's own errors.
In a mock dialogue, the model insists it must first build a mistake-catcher (built by another model, since it can't police itself), then celebrates three hours later that the catcher caught it hallucinating — and proposes building more tooling to fix the mistake. The author's verdict: 'I could have done this myself,' then deletes everything.
Related event: Developers Complain AI Agent Work Is Mostly Guardrails and Tooling(2 posts)→
More from coding & agent
- Submitting a PR Isn't a Flex Anymore in the AI Coding Era — pjausovec · 2026-09-02
- Open-source 7-layer production agentic AI system hits 937 GitHub stars — mdancho84 · 2026-09-02
- Jest creator Nakazawa: six engineering values that matter more as coding agents speed up — bibryam · 2026-09-02
- ODS turns your PC into a private AI server with Ollama, n8n and ComfyUI pre-wired — tom_doerr · 2026-09-02
- Reverse-engineering ChatGPT's memory: 4 layers, no vector DB, no RAG — HowDevelop · 2026-09-02
- Google's $300 credit broke Gemini nodes in ComfyUI, so one dev built a Vertex AI workaround — Cute-Appointment6874 · 2026-09-02