Senticit Meridian Atlas™
An AI policy that lives in a slide deck governs nothing. Meridian Atlas™ is the workspace where AI policies are authored, versioned and operationalized — each one scored for whether it is specific enough to follow, checked for whether anything downstream actually enforces it, and tracked as it is amended so a policy that has quietly drifted is visible.
What it reports
- Policy register
- every policy, with status and version
- Governance coverage
- how much of the estate a published policy covers
- Policy quality
- scored per policy, and as an estate mean
- Enforcement
- which policies name a control that exists
- Amendment read
- what the last revision actually changed
What Atlas does
Policies measured, not just filed
Atlas asks the questions a policy library usually cannot answer: is this specific enough to act on, does it name someone accountable, is anything actually enforcing it, and what did the last revision change?
Author & Version
Policies move through draft, published and retired with their full version history intact. A policy at version four with nothing on file is three revisions nobody can see, and Atlas says so rather than showing you the latest and calling it the record.
Policy Quality Scoring
Each policy is scored on whether it is specific, names an accountable owner, and states what happens when it is breached. "We take AI safety seriously" scores exactly as well as it deserves to.
Governance Coverage
What share of your AI estate is actually covered by a published policy — kept separate from quality rather than blended into one figure, because a small set of excellent policies and broad mediocre coverage are different problems.
Linked Controls, Actually Checked
A policy can name the control that enforces it. Atlas checks that the control exists — a policy pointing at nothing is reported unenforced, which is the gap that audits find and policy libraries do not.
Amendment Read
What the most recent revision changed, per policy — the only reading in the workspace that looks backwards, and the one that catches a policy weakened by a quiet edit.
Meridian Suite Integration
Atlas files every review to the shared Evidence Room. It contributes evidence to the Senticit Intelligence Score rather than a score of its own — policy governance is two orthogonal readings, coverage and quality, and the suite reports both rather than averaging them into a number that means neither.
Atlas is a standalone policy workspace today, by design. The handshake in which Compass scores and Sentinel enforces guardrails authored in Atlas is planned, not built — Sentinel's runtime ethics policy currently comes from Compass. We would rather say that plainly than imply an integration you would only discover was absent after buying it.
Why it matters
The benefits of Meridian Atlas™
- Policies scored for whether they can actually be followed
- Coverage and quality reported separately, never blended
- Policies naming controls that do not exist, surfaced
- Full version history, with nothing quietly overwritten
- A backward-looking read that catches weakened policies
- Every review filed as evidence, not just stored
Find out what your AI policies would score
Send us your current AI policy set and we will show you what Meridian Atlas™ reports — coverage, quality, and which policies name a control that nothing implements.