Team

Productivity

Mithril Medicine had one full-time employee supported by outside specialists and laboratories; its acquisition by Anthropic is described as a case study in how Claude can help coordinate drug research, with clinical progress still to be tested.

Tsinghua, Zhipu and partners introduced HarnessPro, which lets agents study their own runs and rewrite their harness without reference answers. The team reports a 2-point pass-rate gain and a 26.1% cut in cost per task, part of a broader Chinese push into recursive self-improvement around model scaffolding rather than weights.

Research • Pandaily

A Google-led research team built StreamVibe, a video agent designed to watch people cook or assemble furniture and speak up only when help is needed. A benchmark shows none of the tested models reliably pick the right moment in longer videos, as errors compound and response quality drops sharply.

Research • The Decoder

Researchers at Y Combinator and Olam Labs tested AI agent swarms in Minecraft, unleashing hundreds of agents to collaborate or compete. The 50-agent teams failed to build a thriving civilization, falling into paralyzing bureaucracy, offering insights into real-world multi-agent coordination challenges.

Research • Hacker News