Google Research's harness² Applies RSI Techniques to the Agent Harness with a Fixed Model
LangChain · x · 2026-10-19
Google Research released harness², techniques for running RSI-like self-improvement on the agent harness while keeping the underlying model fixed — a process they call "harnessing".
Sydney Runkle (LangChain) notes this is essentially the same as hill climbing or continual learning, and shares more details in a thread.
More from coding & agent
- DecisionAct: open-source Python agent framework using decision models instead of LLM replies — AgitatedReveal07 · 2026-10-19
- 2-GPU R9700 vLLM rig hard power-offs: tensor-parallel all-reduce transient, not wattage, kills 1200W PSUs — brainchillzZ · 2026-10-19
- Agent Status Protocol: A Framework-Agnostic Way for AI Agents to Report State — kalyan_kpl · 2026-10-19
- Nadella: Windows being redesigned as a safer OS to run AI agents — wateriscoding · 2026-10-19
- Krea2 vs MOSS vs Qwen2.1 vs Flux3 hands-on: Flux3 flops hard — CaptainBackwoods447 · 2026-10-19
- AI performance engineering meetup: GPU rooflines, agent trace-to-training loop, Homa transport — cfregly · 2026-10-19