← All work
PRODUCTIZEDGovernanceMCPCompliance

AI Governance Suite

Audit-ready AI governance as working software, not slide decks

Everything you need to prove your AI is under control: risk assessments, policies, and training records mapped to federal requirements. Ready when the auditors ask.

8

Governance MCP tools

6

Framework crosswalks

5

Regulated verticals covered

SHA-256 chained

Evidence integrity

The problem

EU AI Act enforcement and OMB M-25-21 mandates are live, fines reach 3% of global revenue, and 91% of mid-market companies use GenAI — but enterprise governance platforms start at $100K/year. Mid-market and public-sector teams are priced out of compliance.

What we built

01

Built GovernanceIQ as an MCP server: ask Claude to assess an AI use case and it scores risk, classifies high-impact AI under M-25-21, identifies required controls, and drafts an audit-ready compliance plan in minutes.

02

Encoded the regulatory knowledge as data, not prose: crosswalk tables mapping NIST AI RMF controls to EU AI Act articles, ISO 42001 clauses, CMMC, HIPAA, and FINRA requirements — queryable from a CLI or MCP tool.

03

Designed Compliance Literacy around evidence: every cohort produces a hash-chained manifest (rosters, role calibration, quiz scores, regulatory mappings) that an auditor can verify artifact-by-artifact.

04

Packaged all three as engagement-driven products with defined deliverables — the anti-pattern to open-ended consulting.

Architecture

GovernanceIQFastMCP server: risk engine, compliance gap engine, lineage engine, report engine
CrosswalksData-driven NIST RMF ↔ EU AI Act ↔ ISO 42001 ↔ CMMC ↔ HIPAA ↔ FINRA mappings
Risk registerStructured likelihood/impact scoring with versioned YAML state
Evidence packagesSHA-256 hash-chained manifests, branded audit PDFs, recertification tracking

Outcomes

  • M-25-21 use-case assessment compressed from weeks of consulting to a minutes-long, repeatable AI workflow
  • One query returns a control's obligations across six frameworks simultaneously — the crosswalk auditors actually want
  • Training evidence that survives an audit: signed, hashed, and mapped to specific regulatory clauses
  • Informed by lived governance experience: Vice Chair of a federal AI Governance Board, former Chief AI Officer

Stack

PythonFastMCPTyperFastAPIYAML/JSON schemasPostgreSQLNeo4j (lineage)PDF generation