Turnpanel Launches a Local-First AI Workspace for Computer-Operating Agents

Summary

Turnpanel, created by echowin founder Kaushal, is presented as a local-first workspace designed to reduce the need to move context among AI chats, coding agents, terminals, browsers, files, and scripts. An agent can operate across the user’s computer, orchestrate other agents and tools, and retain context over time. Background maintenance processes convert frequently used knowledge and tasks into reusable memories and skills stored locally; the workspace can become more useful through completed tasks, but agents cannot change their own permissions. The product includes a visible browser with DOM and CDP access, and it benchmarks a machine to recommend and configure suitable local models. Users can also connect Codex, OpenRouter, OpenCode, or another OpenAI-compatible provider, switch models for different jobs, and assign different models to helper agents. Turnpanel can monitor conversational harnesses such as KOD and Codex, keep their terminals visible, and answer simple questions using existing workspace context. Its mini-app SDK supports persistent dashboards, review queues, and other small utilities without requiring a new project and deployment stack. Work is organized into Spaces that limit which memories, files, services, browser sessions, models, and actions are available to each job, and agents cannot expand those grants. Logins, payments, destructive actions, and follow-up confirmations return control to the user. Complete workspaces, including jobs, terminals, browser streams, applications, and long-running processes, can be accessed remotely through a browser or PWA. The product is in public beta for Mac and PC, with Linux in development. A basic local version is free, while the Pro tier adds remote access and higher hosted-service limits.