Understand
Fiducia OS.

Overview

Fiducia OS is the governance layer for the agentic enterprise. Its production offering, Assurance, is an AI-native IT risk management and governance platform built for the second and third lines of defense at regulated financial institutions. It ingests your frameworks, policies, and live operational data, maps controls to risks continuously, tests controls across the full population, and generates audit-ready evidence, replacing manual spreadsheets and point-in-time sampling.

Governance platforms ask whether AI is safe to deploy. Agent security asks whether an agent is compromised. Fiducia asks whether an autonomous action stayed within enterprise authority, risk appetite, and regulatory obligation, and produces the evidence that answers it.

The three products

Three products sit on one shared substrate. Assurance is in production and carries everything documented here. Provenance and Runtime are in development; their status is stated wherever they appear.

How it works

Every finding travels one path, and that path is the platform's central governance control. Generation and evaluation are kept structurally separate, and no AI-generated finding, score, or recommendation reaches a human reviewer without passing structured evaluation first.

1. 01Agent

   Assembles context, reasons, and drafts the finding.

2. 02LLM-as-Judge

   Scores against calibrated rubrics. It never generates findings.

3. 03Consensus

   Independent agents must converge, or the finding escalates.

4. 04Human in the loop

   A named reviewer validates, modifies, or rejects. The decision of record.

5. 05Ledger

   Append-only and lineage-traced, exportable to your SIEM.

Where humans are mandatory

Human-in-the-loop is a designed control, not a fallback. Agents recommend; named people decide.

• All critical and high severity findings require human validation before escalation or reporting.
• Any compliance score below a configurable threshold triggers human review.
• Agent-recommended remediation actions require human approval before execution.
• Consensus disagreements route to qualified human reviewers.
• Any action evaluated as outside a delegated authority limit routes to the named accountable human.

The platform

Assurance runs six tiers, each operated by a specialized agent with expert humans in the loop. Most platforms stop at the first tier, an inventory or a system of record. Fiducia runs all six as one connected fabric, which is what lets a finding carry its obligation, its authority, its evidence, and its lineage in a single object.

The shared substrate

See the plane-by-plane view on the platform overview.

Determinism for regulated environments

Regulators require reproducibility. A score of 78 out of 100 must mean the same thing whether it was generated by the system or assessed by a human expert. Fiducia resolves the tension between stochastic models and a deterministic mandate by wrapping the model in a governance framework, not by constraining it into uselessness.

Authority evaluation is deterministic and involves no model at all.

Integrations

The Integration Fabric provides more than twenty native connectors on a connect-once, govern-everything model. Assurance reads the systems you already run, produces the independent challenge evidence your second line stands behind, and pushes findings and evidence out to the tools you report from.

Deployment options

Workspace & roles

What a user sees is a workspace of linked registers, each a view onto the same Governance Graph. Six personas carry pre-defined permissions and scope to domains, so a steward for one domain sees that domain and nothing else.

Personas

Coverage

Coverage is maintained in bands, and a framework is listed only where its obligations are parsed to clause level and mapped to controls, so a control you enforce is already tied to the obligation it satisfies.

Browse the mapped rules on Solutions.

API

Everything available in the console is reachable through a REST API with scoped, member-bound service tokens, which is how most institutions wire Assurance into existing GRC and reporting tooling. Tokens are self-service for the Product Owner persona. Writes are limited to platform records; there is no API path that writes to a connected source system, by design.

List endpoints are cursor-paginated, responses are rate-limited per token, and every response carries a request id to quote in a support case. Full endpoint schemas and sandbox access are available from your account team.

Core concepts

The lexicon: the words the platform uses, defined once and used consistently.

The SeamWhere data becomes a decision.

The join between the data estate and the AI estate. Both run well alone. The unmanaged place between them is where AI failures actually happen, and today no one owns it.

The EstateEverything that can read data or act.

Datasets, models, copilots, agents, service accounts, and third parties. If it can read or act, it is in the estate, and it is in scope.

The Consequence GapThe time between an action and its effect.

Old controls all depended on this gap being wide enough for a person to step in. Autonomy closes it to zero, which is why controls that review after the fact stop working.

Standing AuthorityWhat an agent may do while idle.

The real surface of autonomy is not what an agent is doing now, it is what it is allowed to do at any moment. Standing authority is the access that sits there waiting to be used or abused.

The Authority LadderFive rungs of what a system may do.

From read only, to suggest, to act within reversible bounds, to consequential action, to self directed inside a mandate. Every step up is declared and recorded, never a default.

Shadow AutonomyTools quietly gaining the power to act.

A copilot that starts writing back. A workflow that starts deciding. Authority no one granted on purpose, accumulating faster than any audit can catch.

The Attribution TripleWhich agent, for whom, under what authority.

Every action answers all three or it does not commit. It is the minimum record that makes an autonomous action accountable to a person and a policy.

Blast RadiusHow far one confused agent can reach.

Least privilege is not hygiene here. It is the containment boundary that decides whether one compromised agent touches one account or ten thousand.

The Trust GateNothing reaches a model until it clears.

A single checkpoint between the data estate and any model, asking three questions: is it sensitive, is it sound, do we know where it came from.

The Kill SwitchStop one action, or everything, at once.

Immediate, total, and tested. An agent that cannot be stopped is an agent that cannot be deployed. Not a feature. The precondition for deployment.

Living EvidenceProof that writes itself.

The audit trail is a byproduct of governed execution, current the moment anyone asks. Not a binder assembled for an exam, and never stale.

Proof on DemandThe exam becomes a query.

Ask the system a question a regulator would ask, and the record answers with the dossier and the evidence attached.

The Governance GraphThe live map that ties it all together.

Every dataset, model, agent, control, and rule, connected. Pull one node and see everything it touches. Impact analysis becomes a gesture instead of a project.

The Examination SurfaceEvery question an examiner can ask.

The full set of questions a regulator could put to you, and whether you can answer each one right now. Fiducia shrinks the unanswered part of that surface to nothing.

Frequently asked

Need help?

Visit Support for channels, how cases are classified, and what to send, or book a working session to walk it against your own estate.