For security teams
Approve the routine, not the model.
Your people already connect AI to email, CRM and tickets. Cresec gives you an answer besides yes and no: a tool whose every action is fixed in code.
What you review
Review the apps and actions each signed tool version can reach, and keep access within your workspace.
What you get
- A list of the routines turned into tools: owner and systems touched.
- The tool holds no app or model keys; the gateway uses the person’s connection, or the workspace’s model key, for an approved call.
- Drafts by default. Sending requires a send grant for that person and tool; a write grant alone does not allow it.
- Signed records identify the person and tool version behind calls routed through Cresec.
How it fits your stack
Cresec uses OIDC for company sign-in. We configure and check it against your identity provider, its settings and consent, during setup. Each person connects their own app accounts, under the app’s usual consent. Cresec is designed so a tool runs on Cresec or on your own infrastructure, with the same identity, gateway and per-person access model; neither is deployed for customers yet. If the control service is unavailable, calls stop with an error.
Questions your team will ask
Does the model ever hold credentials?
No. The gateway uses the person’s connection, or the workspace’s model key, for an approved call.
What if the model is fooled?
It can change words in a draft. It can’t change the action, the recipient or the records read.
What about agents that plan their own steps?
Out of scope.
Bring one routine to a 30-minute review.
We’ll attack it in front of you.