AI Managers are the always-on layer: department heads that never stop reading their domain.
The roster
- ›Product Manager AI - watches discovery, requirements, and roadmap health in Product OS.
- ›Project Manager AI - watches delivery: sprints, workload, slippage, and blockers.
- ›Account & Delivery AI - watches client work and account health across delivery and customer success.
- ›Cortex Executive AI - the cross-company manager: goals, strategy drift, and the weekly operating rhythm.
- ›Employee Copilot - the individual's manager: your tasks, your calendar, your follow-ups.
What a manager run produces
On each pass, a manager reviews its domain and produces findings and proposals: "sprint 14 is tracking 20% under velocity; recommend moving these two items", "this account's health dropped after two missed check-ins". Findings are informational; proposals are actions that respect the approvals model.
Configuring them
Command Center, AI Managers shows each manager, its cadence, and its recent runs. You set autonomy per manager - most teams run managers in suggest mode permanently and treat their proposals as a curated to-do list, delegating only narrow verbs (like sending internal reminders) after trust is established.
Managers vs your agents
Managers cover the standing concerns every company shares. Custom agents cover the workflows unique to yours. If a manager already watches something, do not build an agent for it - tune the manager's autonomy instead.
Tip: The Employee Copilot is the gentlest introduction to AI Managers for a team: it only proposes things about your own work, so there is nothing organizational to approve.