InAgent achieves 90.2% on OSWorld: the computer-use agent gap narrows for the Chinese stack
A Chinese CUA becomes the first to surpass 90% on OSWorld. The gap with US frontier labs is now a harness gap, not a model gap.
5 h ago 5 6
Beyond grep: the thesis of the rich-context AI coding harness
Ars Technica states what code agent practitioners were observing: context strategy has become the real lever - not the underlying model.
Jul 20, 2026 at 16:42 9 12
QCon AI Boston: "prompts → platforms, harnesses, evals" - the field validates the thesis
The QCon AI Boston conference captures the turning point: AI production has shifted from a matter of prompts to a matter of platforms, harnesses, and evaluations. A convergence, not a trend.
Jul 17, 2026 at 14:19 16 8
Anthropic provides its user manual for scaling agents: layers, tokens, reliable execution
Two engineers from Anthropic publish field lessons: for an agent that needs to scale, it's no longer the model that limits, but the harness. Three pillars: layers, budget, execution.
Jul 15, 2026 at 12:34 30 5