Senticit Meridian — Intelligence Score dashboard

Senticit Meridian™ / Suite
Connecting…

Live bearing

A single bearing on AI trust.

Ten instruments, one reading. Every panel below is a live evaluation over the Meridian API, rolled into one board-facing number — the Senticit Intelligence Score.

Assessed organization

Press Delete to remove this organization.

Senticit Intelligence Score

—
/ 100
—

Equal-weight average of the instruments that score, computed by POST /api/sis. A blocking finding caps the status, whatever the average says.

Roll-up · live from the API

How the number is assembled

POST /api/sis
Evaluating…
◆ Narrated · Signal opportunity headline · deterministic fallback

Slant · Ledger™ · Pulse · Atlas contribute qualitative evidence, not a headline score — they inform the record without moving the number.

The constellation · all ten live

Ten products along one pipeline

POST /api/{product}/evaluate
Discover Assess Prepare Approve Monitor Prove

Drill-in · one instrument at a time

Sight an instrument to read it

Select any instrument in the constellation above — its drill-in opens here.

Radar — Readiness & Peer Benchmark

Discover/Assess
POST /api/radar/evaluate
Overall readiness
—

Top control gaps

Remediation roadmap

Signal — Opportunity Brief

Discover · risk-adjusted upside
POST /api/signal/evaluate
— —
Risk-adjusted upside
—
—
◆ Narrated headline · opportunity brief

Recommended next steps

Tether — Agent Fleet

Discover · who is running, and has it drifted
POST /api/tether/evaluate
— agents observed
Fleet verdict

Unregistered agents

Drift findings

Highest-risk agents

Slant — Fairness, Drift, Explainability & Accreditation

Approve · clinical AI
POST /api/slant/evaluate
Explainability verdict
—

Method catalog · enriched findings

Compass — Ethics Review & Approval

Approve · value alignment
POST /api/compass/evaluate
Ethics score
—
Approval decision
—

Policy alignment

Sentinel — Integrity Trace & Ethics Gate

Monitor · runtime integrity
POST /api/sentinel/evaluate
— —
Integrity / 100

Ledger™ — Continuous Control Monitoring

Monitor · evidence freshness
POST /api/ledger/evaluate

Pulse — Executive Risk Telemetry

Monitor · board-ready RAG
POST /api/pulse/evaluate
— —
Risks worsening
—

Top risks

Atlas — Policy Workspace

Prove · operationalize policy
POST /api/atlas/evaluate
— policies

Policy register

Kept policies — versioned, shared across your organization

Marshal — Workflow Governance

Prove · what the workflow did, and who approved it
POST /api/marshal/evaluate
— steps

Approval checkpoints

Replayable chain

Step verdicts

Trends · score history

How scores move over time

Points are real evaluations — captured on demand, and automatically when an org is created or edited.

Captured Why SIS Per-product
Organization Points First Latest Δ

Evidence Room · persisted & filterable

Every evaluation is filed and hashed

Supabase · evidence_records · RLS
Product Score Kind Ref Tags Collected Content hash

Read live from evidence_records via GET /api/evidence, filtered by organization. Each evaluation is filed server-side (keyed by product, org & subject) under deny-by-default RLS when a Supabase service-role key is set. Select any row to inspect the full record — hash, keying & provenance.

Remediation · derived from the live evaluation

Turn findings into tracked fixes

Supabase · remediation_items · RLS
Evaluating…

Recommendations are regenerated from the current scores every run — products below the remediation line, blocked Ethics-Gate verdicts, and a weak or critical SIS. Move an item through open → in progress → done, or file it as an OpenProject work package. Status and the filed work-package link persist tenant-scoped under RLS (remediation_items), or per-browser when no server store is configured. Filing uses a server-side OpenProject REST integration when creds are set; otherwise it copies a ready-to-paste task.

Calibration · scoring health

Every product's scoring, validated

—

Live from GET /api/calibration — a quality-knob sweep per product checks the score is monotonic (better inputs never lower it), in range, and band-discriminating. No ground-truth labels; sensitivity shows each input's contribution to the score.

Meridian v1.0.0 · 37facc0f