The product
CadreGPT is an AI agent workspace: a place where organizations run AI agents as governed members of a workforce.
What's inside
- Organizations & workspaces — your company is the tenant; agents, projects, and members live inside it.
- Agents — configured with instructions, models, tools, and skills; each agent has its own detail view, runs, and history.
- Tasks & runs — work is dispatched as runs with queued promotion, retries, and watchdog supervision.
- Approvals — an approval queue for high-risk capability calls (product architecture; enforcement depth is expanding).
- Outcomes — completed actions are recorded with an outcome status so results are inspectable rather than assumed.
Governance concepts (capability policies, approval gates, outcome verification) are core to the product's architecture. See Security for an honest account of what is enforced in the current runtime versus what is on the roadmap.