Term: Copilot
~5 min read
Estimated time: ~5 min read — for the in-app brief plus opening the primary source.
What this is
A copilot is an AI assistant embedded in a workflow to help a human work faster — drafts, suggestions, lookups — while the human remains accountable.
Everyday example
Microsoft 365 Copilot drafting an email you still send, or a CRM assistant suggesting next steps you approve — human remains the pilot.
A copilot helps you work — draft, suggest, summarize — while you stay in the chair.
- GitHub Copilot, Microsoft 365 Copilot, and in-app assistants are this pattern.
- You review and send. The copilot does not run the process alone.
- Risk is still real: leaked data, weak drafts, over-trust.
- Different from an agent, which takes multi-step actions.
Next action: Name one copilot in use and write who reviews before anything leaves the building.
What changes in how you lead
How decision rights, process, and ownership should change.
- Copilot licenses are not a strategy until review rules exist.
- People still own the sent email, the filed brief, and the merged code.
Compare related ideas
Copilot vs Agent / agentic AI
A copilot suggests or drafts; you click send. An agent can plan and take multi-step actions across systems (create tickets, send mail, update records) — higher leverage and higher blast radius.
Open Agent / agentic AIDeep dive
Good copilots show sources, allow edit, and fit existing tools (Office, CRM, HRIS).
Bad deployments hide automation inside decisions people still 'own' without understanding.
ChatGPT, Claude, and Grok can act as general copilots; enterprise copilots add company data connectors and admin controls.