Agents and harnesses · runtimes-frameworks
Microsoft open-sourced an evaluation framework for testing enterprise AI-agent behavior.
Microsoft open-sourced an AI evaluation framework designed to test the behavior of enterprise AI agents.
Read the original at infoworld.comOpens the publisher's site in a new tabMore in Agents and harnesses
Anthropic announced Claude blueprints for retailers to build shopping and merchant agents ahead of the holiday season.Sep 2China released a national standard for interconnected AI agents, with 50 companies reportedly piloting it.Sep 3OpenClaw releases version 2.0 with a rebuilt browser app, simpler setup, shared sessions, and revised security controls.Aug 31NVIDIA releases PAIR, an open-source Personal AI Router that pools idle home-computer GPUs for local inference.Sep 3Anthropic revises Claude Code’s permanent weekly usage limits effective September 14.Sep 4