AgentRun harness shifts LLM steps to code, cutting compliance workload cost by 90%
garrytan · x · 2026-09-22
- TypesafeAI built AgentRun, a harness on top of their model Jev, targeting repeatable knowledge work customers run thousands of times a day.
- Core mechanism: the harness learns the job as it runs, moving steps from LLM calls to deterministic code.
- Cost: 100,000 compliance alerts cost >$290K on Opus 5; with agentrun() it drops to <$26K, 90% savings.
- The team's writeup argues that focusing on one use case — repetitive work at scale — enabled a fundamentally different harness design.
More from coding & agent
- Jev-style models called a major unlock for fast, cheap browser agents — multiply_matrix · 2026-09-22
- Generating responsive UIs in under 2 seconds with shadcn plus Mobbin MCP — msharmas · 2026-09-22
- Nat Friedman admits Meta's Muse was inspired by OpenClaw, bought hundreds of Mac minis for the team — EdwardSun0909 · 2026-09-22
- SkillLift cuts agent skill-evolution token cost 40-70% by ranking, not rollouts — dair_ai · 2026-09-22
- Deel launches Akai: 8,000 agents doing work of ~600 employees, added $140M ARR in 90 days — AIwithGhotai · 2026-09-22
- exe.dev wins over developers: SSH into root VMs, plus the underrated Shelley coding agent — davidcrawshaw · 2026-09-22