Use cases
Fourteen ways this shows up in a real estate.
The same control plane, entered from wherever the problem already sits — a persona who owns it, a risk you are trying to prevent, a platform the agent already runs on, or a runtime where the decision has to be made. Every page states what it enforces and what it does not.
START WHERE YOU ALREADY ARE
Four ways in.
These are entry points, not product tiers. They describe the same enforcement point from four angles, because the person asking is rarely the same twice.
By persona
Who owns the problem
By risk
What you are trying to prevent
By platform
Where the agent already runs
By cloud and runtime
Where the decision is made
WHO THIS IS FOR
Designed for the teams accountable when AI takes action.
Three buyers, one governed action trail. Each sees the same decision from the side they answer for.
Per-call authorization, injection defences, data controls and stop controls — at supported enforcement points.
Security → GRCEvidence chains and control mapping for governed actions, in the form a reviewer actually asks for.
GRC → Platform & AI EngineeringA policy and evidence layer around the agents, models, frameworks and tools you already run.
Platform & AI →BEFORE ANY OF THEM
Every case here governs an agent that is already running.
That is the half most products stop at. The other half is assurance — the gate before production: an immutable snapshot, bars set before the run, and distinct exit codes a deploy step can consume rather than annotate afterwards (shown today as a demo, not yet the automated stop in your pipeline).
A governance tool that reports a pass it cannot support is worse than one that reports nothing, so a domain the environment cannot measure returns not_assessed with the reason.