Offloop launches AI Agent Workspace for recurrent work
Offloop’s private beta centers recurring agent work around persistent Channels that can be triggered by schedules or incoming signals.
The workspace keeps people, agents, files, messages, permissions, progress, and finished artifacts in one shared place. Agents can resume work with prior context, and users can inspect, redirect, approve, or stop runs while they are visible. Offloop says its D1 dispatcher decides which agent acts when multiple agents share a Channel. The company also claims benchmark scores of 84.9% on GDPval and 67.2% on JobBench, at about one-fifth the cost per task of the baselines it tested against. TestingCatalog's note
The workspace keeps people, agents, files, messages, permissions, progress, and finished artifacts in one shared place. Agents can resume work with prior context, and users can inspect, redirect, approve, or stop runs while they are visible. Offloop says its D1 dispatcher decides which agent acts when multiple agents share a Channel. The company also claims benchmark scores of 84.9% on GDPval and 67.2% on JobBench, at about one-fifth the cost per task of the baselines it tested against. TestingCatalog's note
score 5