Origi Framework

Calibrated trust, not maximal trust

The goal is reliance matched to a system's actual, demonstrated trustworthiness — because over-reliance and under-reliance are both failures. A dependable system nobody can interpret invites under-reliance; a polished interface over an undependable system invites dangerous over-reliance. Origi is built to deliver, and to demonstrate, both sides: objective dependability and the affordances that let you perceive it accurately.

Every trust claim is an assurance case — claim, argument, evidence — bound to exactly one policy with an explicit threshold and owner. A claim is satisfied only when the measured outcome meets the threshold and the evidence chain validates: every signature verifies, no hash link is broken, no evidence has expired. Never on a passing number alone.

Nine components of trustworthiness

Each carries its own policy, claim, measurable outcomes, and evidence chain — and each maps to the NIST AI RMF trustworthiness characteristics, so the dossier cross-walks to recognised governance.

The Trust Harness: never trust, always verify

The same claims, policies, and provenance move to inference time. The harness is a reference monitor for trust: nothing an agent produces — plans, messages, tool calls, outputs — is trusted by default. A checkpoint sits at every boundary, and delegated authority can never exceed the authority that delegated it.

Promotion slow.
Demotion instant.

Trust zones set the maximum capability an agent may even request, and the envelope expands only as trustworthiness is demonstrated — every action still individually verified. Promotion requires clearing a higher bar and sustaining it; a single breach or detected attack revokes capability immediately.

The control plane itself is split three ways — an Observer assesses evidence it did not produce, a Judge rules against zone policy, and a Controller acts only on a signed verdict. No component can both decide an agent deserves more power and grant it — the self-escalation path is removed by construction, and every transition is signed provenance.

Don't ask to be trusted.
Show, claim by claim, exactly why you can be.

The framework is itself tested: an independent adversarial program attempts to defeat the controls round after round, every finding becomes a fix and a permanent regression test, and because every run is signed, replayable provenance, the evaluation itself is reproducible. Origi is the trust harness inside Northwind, Lamassu, and the ECCC engagement — the same machinery, proven in the field.

Put Origi around your agents

Book a walkthrough and watch a request cross the harness — pre-flight to post-flight, with the trust dossier it leaves behind.