The model proposes, a person commits, and every call is logged
A safety gate so the AI can suggest actions but never takes a risky one on its own.
Fail-closed: deny unless explicitly permitted
Checkpoint / resume with human approval
The challenge
An AI that can take actions is a risk in a regulated business. You can't ship something where the model decides on its own to submit, close, or commit anything that has consequences.
How we solved it
Every action the AI suggests goes through a policy gate. If no rule allows it, it's denied, and every decision is logged. Anything with real consequences doesn't run at all. The workflow pauses, shows what's waiting, and only a human approval starts it again, picking up right where it stopped, even days later.
The result
The model proposes, a person commits. Nothing important happens on the AI's say-so, and there's a clear record of who decided what, and when.
Technologies
.NETMicrosoft Agent FrameworkFail-closed policy gateHuman-in-the-loopAudit logging