The Harness Singularity: Inside Weco AI's Breakthrough in Self-Improving Agents

Weco AI CEO Zhengyao Jiang said in a Machine Learning Street Talk feature that the company's AID system, which applies recursive self-improvement to agentic scaffolding rather than neural weights, produced seven accepted OpenAI records along with self-evolved anti-cheating mechanisms. The autonomous harnesses generated code Jiang described as alien yet highly generalizable, and he said they outperformed human R&D teams.

In a landmark Machine Learning Street Talk feature, Weco AI CEO Zhengyao Jiang unveils how autonomous harnesses are outperforming human R&D teams. By shifting recursive self-improvement from neural weights to agentic scaffolding, Weco's AID system generated alien yet highly generalizable code—achieving seven accepted OpenAI records and introducing self-evolved anti-cheating mechanisms.