Governance

Diagnostic

A short, structured conversation. We map the AI surface area in your organization: who is using what, where decisions are happening, and what is already exposed. The output is a written read of where you are, what is at risk, and whether an Assessment is warranted. If it is not, we tell you. The Diagnostic is fixed-fee and time-boxed.

Assurance Assessment

A formal, structured evaluation of where AI creates risk across your organization. Each area we examine is scored, and findings past a defined escalation threshold are flagged for mandatory action. The Assessment is delivered as a written report with explicit recommendations and, where applicable, a scoped Governance Build proposal. It is decision-ready evidence: a document that can be read by your board, your insurer, your auditor, and your counsel. Additional workflows, business units, or entities are scoped in writing at signing.

No Governance Build is offered without an Assessment.

Governance Build

Tightly scoped build of the governance system: policy layer, workflow controls, ownership structure, approval and escalation paths, and monitoring mechanisms. We do not write code, configure tools, or run pilots. We define the operating model and put the controls in place that make it run. The proposal you sign defines exactly what you are getting and what you are not.

Governance Build is scoped by whether the AI acts on its own. Which band applies is determined by what the Assessment finds, not selected from a menu. Agentic decision objects are counted across the whole engagement, not per workflow.

Non-Agentic

For AI that produces output a person then acts on: drafting, analysis, classification, recommendation, and summarization. The control layer governs what the system may be used for, who owns the decision that follows from it, what validates the output before it becomes action, and what gets recorded.

Agentic

For AI that takes action itself: systems that execute, transact, or trigger downstream steps with no person in between. Everything in the Non-Agentic tier, plus what delegated authority requires: defined action boundaries, pre-authorization thresholds, intervention and stop conditions, escalation that fires without a human starting it, and an audit trail that reconstructs what the system did and under whose authority it did it.

Agentic AI does not remove the decision. It moves the decision earlier, into the authority you granted before the system ever ran. That authority is the thing this tier governs.

AI Sustainment Program

Ongoing oversight at a defined cadence. Quarterly reviews against the original scoring. New AI use cases assessed against the same scale. Regulatory and operating-environment changes folded into the model. One-year contracts with quarterly cancellation. Standing Cadence and Incident Response is the assurance floor after Governance Build or an Emissaries exit. Standing Cadence is advisory only, for Deployment Blueprint clients who self-implement where that route is permitted. The AI Sustainment Program is where the discipline compounds and where most of the long-term value of working with us lives. The operating model is designed to hold whether or not we are in the room.