Fintra Trust feature
One versioned rulebook for every agent
Fintra Trust, AI Constitution
Example figures- RULES
- defined once
- versioned
- APPLIES TO
- every agent
- consistently
- CHANGES
- tracked
- auditable
- Never move money without approval
- enforced everywhere
- No cross-tenant data access
- enforced everywhere
- Sensitive writes require step-up
- enforced everywhere
- Policy version
- tracked + diffable
Policy in one place, not a hundred prompts
When each agent carries its own rules in its own prompt, policy drifts and no one can say what the organization actually allows. The AI Constitution centralizes it: a versioned rulebook every agent decision is evaluated against, so the same rule means the same thing everywhere and a change applies everywhere at once.
Questions people ask
What is the AI Constitution?
It is a single, versioned set of rules your organization defines once and every agent decision is checked against, so AI policy is consistent, centralized, and auditable rather than scattered across prompts.
Why not just put rules in each agent’s prompt?
Per-prompt rules drift apart and can’t be audited as a whole. A central constitution means one rule applies everywhere, and changing it changes behavior everywhere at once.
Is it auditable?
Yes. The constitution is versioned and diffable, so you can see what the rules were at any point and tie a decision to the policy version that produced it.
How does it relate to per-action decisions?
The constitution defines the rules; the decision fabric applies them to each action and returns a verdict, which is written to the Decision Ledger.
Keep reading
Calculators
One finance briefing a week
New guides, checklists and benchmarks, by email. Unsubscribe from any issue.
Put a decision, with proof, in front of every action
See how Fintra decides allow / step-up / hold / block per action and writes each verdict to a tamper-evident ledger.