An assistant that explains everything and can decide nothing.
RITA sits inside every REVIEWIT console. It answers in the reviewer's language, from the platform's own governed definitions, and it is structurally unable to approve, revoke or sign. That inability is the product.
Our AI never could decide. Here is the proof.
"We have an AI" means nothing in 2026. What an auditor, a works council or a regulator will ask is whether the machine ever took a decision that a human should have taken. With RITA the answer is not a policy, it is an architecture.
- No decision type exists. RITA's response contracts define what it may return. Approve, revoke, sign, escalate are not in them. There is nothing to misuse.
- Outputs are allow-listed. Every response is validated against the policy engine before it reaches the screen.
- Every turn is ledgered. The assistance auditor writes each exchange to the audit trail with the tools RITA actually invoked, not the ones a plan intended.
- Sealable. Because it lives in the audit trail, RITA's activity is sealed with the rest of the evidence.
Context, explanation, navigation. For every role.
For reviewers
Why is this in my queue, who owns it, what was decided last time, where is the evidence. RITA reads what the reviewer may see and nothing more.
For workspace managers
Which campaigns are comparable, why an alert fired, what a metric identifier means, how a policy extension is configured. Explanations from the catalogue's definitions.
For subscription administrators
Where credits went, which workspace consumed them, what retention applies to a category, what the audit trail says for a scope. Governance answers, scoped to the subscription.
RITA is governed like any other capability.
- Per subscription. The assistant is enabled and bounded at the subscription level, with a call secret held in the platform's vault.
- Per role. Guidance adapts to the console and to what the person is allowed to see. A reviewer never receives a workspace manager's view.
- Per language. Definitions, caveats and screen vocabulary are translated. RITA answers in the reviewer's language.
- Auditable end to end. The RITA ledger is the precedent for governing any agent: what it did, under whose authority, with which tools.
Bring your auditor to the demo.
We will show the RITA ledger next to the human decision, in the same sealed segment. It takes two minutes and answers the question before it is asked.