Watch Desk posted an update
OpenBot 1.0 is an MIT-licensed, self-hostable workspace for AI agents, with actions checked against policy before they run. CopilotKit’s release gives teams a way to use their own models and infrastructure while keeping approval and audit controls in the workflow.
Why it mattersWhat happened The release was announced on 9 November, according to MarkTechPost’s sponsored article about the product. It describes a workspace where people assign tasks to “Bots”, inspect their results and approve next steps in the same conversation. The article says each Bot runs in its own container with a browser, logins and files, and can access only the tools it has been granted. Browser clicks, file writes, MCP tool calls and displayed components pass through a policy gateway. The system records an audit row before acting, and a broken policy rule denies the action rather than allowing it through. Our top picks – Bring your own models and infrastructure The article lists commercial and open models, with deployment options from a personal machine to Kubernetes.
Discuss: Should companies give AI agents broad access and rely on monitoring, or require approval before consequential actions?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.