AI engineers and approvals
Work with AI engineers that run security work for your team, give them targets and autonomy budgets, and approve the calls they request.
On this page
Engineers are AI members of your team. Each one has a plan and an agenda, works on authorized targets, keeps notes, and reports findings, all within the autonomy you give it.
What an engineer page shows
| Area | Contains |
|---|---|
| Plan and agenda | What the engineer is working on and what is scheduled, including meetings on its calendar. |
| Authorized targets | The targets it may test. It does not act outside them. |
| Autonomy | Budgets and limits for unattended work. |
| Approvals | Actions waiting for a person to approve. |
| Calls and messages | Conversations with your team and contacts, and received mail. |
| Contacts | People it may contact. |
| Findings | What it has found. |
Watch an engineer’s live work from its worker view.
Approve a call
When an engineer needs to act on your behalf, for example joining a call with access to your account, it asks first. The request opens an approval page that shows exactly what access is requested.
- Open the approval link.
- Check who is signed in and what access the call needs.
- Approve this call or Decline.
- Revoke account access later from the same page if you change your mind.
Engineer permissions are member scopes (staff:*) that an organization owner grants. See Scopes and roles. Safety controls for all engineers are under Operations.