Learn · Path · Governance / risk / compliance reviewer
AI model governance and source review
Four readings + three exercises + three audit workflows. Walks lifecycle, status, benchmark limits, pricing, and source freshness so a reviewer can sign off with a defensible evidence trail — never a certification claim.
Use this path inside a kit
The matching workflow kit bundles this path with the lab playbooks, evaluation prompt sets, and Markdown templates the role needs, and exports the whole thing as a single work document.
Audience
Governance / risk / compliance reviewer
Difficulty
intermediate
Estimated time
57 min · 10 steps
What you walk away with
What you will learn
- How to use lifecycle state as an integration gate.
- How vendor-reported status differs from the independent probe.
- Why benchmark scores are intentionally absent from the catalogue.
- How pricing references age and where the reverification queue surfaces stale ones.
What you will build
- A source freshness checklist scoped to the reviewed provider.
- A lifecycle review note with retirement dates.
- An explicit data-gap list for external follow-up.
- A written external test plan.
- A governance review brief paired with the evidence above.
Evidence artifacts
- Source freshness checklist
- Lifecycle review note
- Data gap list
- External test plan
- Governance review brief
Tools used
Learn hub →
Concept lessons that explain each verified field.
Exercises →
Practical workflows that end with evidence artifacts.
Citation registry →
Every primary-source URL the catalogue references.
Coverage audit →
Per-provider verified-field counts and citation density.
Reverification queue →
Sources due for manual re-check with retrievedAt dates.
Decision brief builder →
Export a Markdown or JSON evidence brief.
Prerequisites
- You can name the provider(s) under review and the regime(s) you are reviewing against.
Timeline
10 steps
Open each step in the order shown. The route on every step is the canonical workspace or lesson — no parallel UI.
- lesson5 min
Model lifecycle: active, deprecated, retired
Read lifecycle as a hard gate on integration risk.
- lesson5 min
Status-aware model selection
Separate vendor-reported status from the independent probe.
- lesson5 min
Why benchmark scores can mislead
Understand why no benchmark score is published in the catalogue.
- lesson5 min
AI model pricing references explained
Read pricing as a sourced reference with a retrievedAt date.
- exercise5 min
Check source freshness and reverification state
End with a checklist of citations that need re-reading.
- exercise5 min
Inspect lifecycle before integration
End with a lifecycle risk note paired with provider deprecation history.
- exercise12 min
Plan an external model test
End with the test plan that pairs your review with workload-specific tests.
- workflow5 min
Walk the reverification queue
Read the queue filtered by provider so stale citations surface for the review.
- workflow5 min
Audit per-provider coverage
Confirm verified-field counts and citation density before sign-off.
- workflow5 min
Open the citation registry
Trace each evidence claim back to its primary source URL.
Start next
Step 1 of 10: Model lifecycle: active, deprecated, retired
Read lifecycle as a hard gate on integration risk.
What this path does not promise
- Compliance certification.
- Risk approval.
- Legal advice.
- Vendor endorsement.
How to use this path
- Open each route in the order shown. The path is a sequenced reading + practice plan, nothing more.
- Keep the artifacts you produce — the shortlist URL, comparison URL, Markdown brief, freshness checklist, or test plan.
- There is no login surface, no progress state, and no completion certificate.
- External workload-specific testing remains your team's responsibility — the catalogue surfaces evidence, not verdicts.
No progress, no accounts, no certificates
- No accounts — the catalogue does not have a login surface.
- No progress tracking — the catalogue does not store which pages you have visited.
- No certificates — the catalogue does not issue completion credentials, badges, or scores.
- Completion is the artifact you produce: a shortlist URL, a comparison URL, a Markdown brief, a freshness checklist, or a written test plan.