AI readiness audit
The Audit
Find out where your AI deployment will fail before something else does. The audit produces a scored report with the evidence cited under every finding, so you can act on the gaps and defend the score on the merits.
Who this is for
- You ship AI in healthcare, finance, insurance, or defense — and one bad model is a regulator's letter, a lawsuit, or a recall.
- You bought or built an AI system and need someone outside the build team to check it before it scales — or before you put it in front of the board.
- The gap between "we have a model" and "we can defend this model" is the gap keeping your leadership up at night.
What you walk away with
- A ~30-page scored report. Every score backed by specific evidence, every recommendation prioritized.
- A 60-minute debrief with your technical and executive teams in the same room.
- A 30-day check-in. We come back to see what moved — and you don't pay extra for it.
Methodology
IntakeWeek 1
Scope definition and non-disclosure agreement execution. We identify stakeholders, define the systems under review, and issue a structured document request covering architecture diagrams, access policies, data lineage maps, and operational runbooks.
Evidence ReviewWeeks 2–3
Deep review of all submitted artifacts. We trace data flows end-to-end, evaluate access control implementations against stated policies, and assess process documentation for completeness and operational viability.
Pillar ScoringWeek 4
Each of the three pillars — data architecture, access control, and process documentation — is scored against defined maturity levels. Every score is backed by specific evidence and documented rationale.
ReportWeek 5
Delivery of the scored written report. Findings include risk flags, gap analysis, and prioritized recommendations for each pillar.
DebriefWeek 6
A 60-minute walkthrough of findings with your stakeholders. Questions addressed, priorities clarified. Followed by a 30-day check-in to assess progress against flagged items.
Scoring rubric
Each pillar is evaluated across four maturity levels. The resulting score is transparent — you see exactly what was measured, what evidence was cited, and where your organization sits on each axis.
What a scored report looks like
The scored output is a triangle on three axes. An equilateral shape means balanced maturity. A lopsided shape tells you which pillar to invest in next — backed by the evidence citations that produced each score.
Engagement terms
- Typical engagement duration: 3–6 weeks depending on scope and organizational complexity.
- Access requirements: architecture documentation, access control policies, data pipeline configurations, stakeholder availability for interviews.
- Non-disclosure agreements executed at intake. All findings remain confidential.
Go deeper
Methodology
The full rubric, in public
Every criterion the audit scores against, at every maturity level — crosswalked to NIST AI RMF, ISO/IEC 42001, the EU AI Act, SR 11-7, OWASP LLM Top 10, and MITRE ATLAS.
Read the methodologyWorked examples
Six AI deployments, scored
Customer-support LLM, fintech fraud model, medical-imaging assist, multi-agent code review, mature LLM, EU AI Act employment screening — each scored across the four pillars with the rationale.
Read the examplesPricing
Engagements sized to scope. Reach out to discuss your organization's requirements.
Not what you need?
Need something built, not audited? Fixed-scope AI engineering with prices on the page.
See what we build