17 researchers led by ex-Huawei Noah's Ark head publish definitive guide to building AI agents
blaizedsouza · x · 2026-10-11
A comprehensive AI agent paper by 17 researchers led by the former head of Huawei's Noah's Ark AI lab is being called the most detailed on agents yet. Core claim: agent quality comes from the model plus the harness around it — the model alone doesn't explain performance.
Key sections:
- p.4: the agent loop in one diagram
- p.5: the full harness, 6 parts from observation to verification
- p.7: four eras from prompts to context to harness to agent-native training
- Section 5: each of the 6 harness parts in detail
- Section 6.3: which harness setup fits which task
- Section 7.3: SWE-bench results by model and harness
Industry context cited: Stripe's coding agents merge 1,000+ PRs weekly, over 70% of Uber's PRs now come from agents, and Spotify has merged 2.5M+ automated maintenance PRs.
More from coding & agent
- Prime Agent orchestrates 2,000+ agent swarm to rewrite itself in Rust, 13x faster startup — GregCook2011 · 2026-10-11
- Dario Amodei predicted software engineering automatable in 12 months — tech leaders say it's coming true — matthew_d_green · 2026-10-11
- Regents Labs runs a daily AI paper digest with ChatGPT Pro, spotlighting Prime Intellect's agent stack — seanwbren · 2026-10-11
- Building an Agent Feedback Loop to Auto-Tune Game Difficulty via Headless Simulation — zimmer550king · 2026-10-11
- Python vs Go vs Rust for production agentic systems: a six-dimension practical evaluation — mostly_deterministic · 2026-10-11
- Agent Use Cases: a free site scanning the web daily for 1,000+ real AI agent use cases — Roger_M_Taylor · 2026-10-11