Anthropic engineer: the longer your agent runs, the costlier its mistakes — build verification in
Roger_M_Taylor · x · 2026-09-18
An Anthropic engineer explains in a 28-minute breakdown how the Claude Code team runs agents that check their own work:
- Core idea: the longer an agent runs, the more expensive its mistakes get — so verification should be built into the artifact itself, not left to human review.
- The talk traces a workflow evolution: Prompting → Loops → Graphs → Self-Verifying Systems, arguing prompting was the old workflow and loop/graph engineering is the next one.
More from coding & agent
- Distributed-Slides: an MCP server that compiles agent-written talks into offline presentations — arthurcolle · 2026-09-18
- Vercel skills CLI adds Notion-hosted agent skills, no Git repo required — ivanhzhao · 2026-09-18
- Vercel teams have long written GTM and on-call skills in Notion — ivanhzhao · 2026-09-18
- eve adds automatic model selection: agents pick models by task difficulty before inference — cramforce · 2026-09-18
- DHH on Lex Fridman: Nov 2024 split coding into two universes, agentic coding is transformative — zakkohane · 2026-09-18
- Vercel CLI now deploys static artifacts in under one second, build step skipped — cramforce · 2026-09-18