From an agent's request to an authorised, evidenced action
Most governance tools stop at the register. Cogna8 keeps a governed view of the facts an agent relies on, checks them against your controls and policy, and decides before the action runs. The same inputs always give the same decision.
What happens when an agent wants to act
- 1
The AI system is known and governed
The agent is registered in the inventory with an owner, a risk tier and the controls that apply. This is the governance record everything else hangs off.
- 2
Context becomes governed state
Facts the agent relies on (amounts, statuses, approvals) are extracted into typed, versioned state items with their source, rather than left as loose text in a prompt.
- 3
Conflicts are detected
When two sources disagree, for example two different claim amounts, Cogna8 records the conflict and its severity instead of silently overwriting one value.
- 4
Controls and policy are evaluated
The proposed action is checked against gate policies derived from your controls: required state, unresolved conflicts, thresholds and who must approve.
- 5
A decision is made before the action
The gate returns allow, needs approval or block. The decision path does not depend on a model judgement, so it can be replayed and audited.
- 6
A receipt becomes evidence
Every decision is written to an append-only trail with what was asked, what was known and which rule applied. Assurance uses these receipts as evidence.
One call before the action
Agents call the gate with the action they want to take. The response says whether it may proceed and why.
{
"session_id": "sess_abc123",
"action": "send_invoice"
}{
"verdict": "BLOCK",
"action": "send_invoice",
"decision_id": "gd_a8f2e1",
"reasons": [
"Unresolved conflict on project.budget",
"Missing required state"
]
}Full reference on the Developers page.
Light entry, deep control
You do not need the gate on day one. Most organisations start with the inventory, risk and controls, then connect the agents whose actions carry real consequences.
- Inventory, risk and controls work with no change to how your AI runs
- Connect one agent at a time to the gate
- Governance records and runtime decisions share one evidence trail
Let's connect
Tell us where your AI programme is today. We start with a short, scoped pilot on your own AI estate, and every enquiry is handled in confidence.