Key notes from a talk on the Codex agent and how harnesses shape model optimization
omarsar0 · x · 2026-09-30
AI researcher Elvis (omarsar0) live-tweets key notes from a technical talk, starting with the Codex agent and its harness. His core point: harnesses inform how the models themselves are optimized — training objectives are tied to the agent's runtime environment. He also poses the question "What comes after loops?", hinting at the next stage of agent architecture beyond the agentic loop.
Related event: Agent harness and loop engineering are reshaping model optimization(2 posts)→
More from coding & agent
- Google ships Jetpack Compose A2UI renderer to turn agent streams into native Android UI — rseroter · 2026-09-30
- v0 now lets ChatGPT Plus and Pro subscribers build with their ChatGPT tokens — tomjohndesign · 2026-09-30
- Teaching agents from outcomes: an async reflection loop that turns resolutions into rules — FirstClothes6582 · 2026-09-30
- Vorflux partners with OpenAI: ChatGPT subscriptions now OAuth into its agent harness — ycombinator · 2026-09-30
- Open Dots: a 4.3k-star open-source, self-hosted alternative to OpenAI's Dots — matchaman11 · 2026-09-30
- ChatGPT subscription now works in 16+ partner products including Devin and Notion — charliermarsh · 2026-09-30