[REGENT]
Governed Engagement Engine · exautomatica.com/regent
Outreach your compliance officer
can sign off on.
can sign off on.
Your AI proposes. Independent rules govern. The audit proves. Every decision — sent, held, or escalated — is on the record before anything reaches a patient or customer.
How it works
01 · Context
Hard stops
Contactability, crisis flag, global opt-out — cleared before AI sees the case.
02 · Judgment
AI proposes
Reason, intent, channel, content. Version-pinned. Integrity-verifiable.
03 · Governance
Rules decide
Consent, content policy, frequency, window — run after the proposal.
04 · Action
Outcome
SEND · HOLD · ESCALATE · DEFER — all traced.
Four outcomes. All on the record.
SENT
req_9f3a2b1c · 14:32:07 UTCLab results ready. All four governance checks passed. Message dispatched.
Hard stops
✓ Clear
AI judgment
SEND
Consent & content
✓ Pass
Frequency cap
✓ Pass
HOLD · Terminal
req_8e2b1a0d · 10:41:08 UTCAI proposed a check-in for Sarah — quiet for six weeks. REGENT held it. Her care team had spoken with her the previous day.
Hard stops
✓ Clear
AI judgment
SEND
Recent human contact
— HOLD
Future outreach
New customer request
ESCALATED
req_7d1a0f9c · 16:47:14 UTCA routine post-discharge engagement was requested. REGENT found a crisis flag before invoking AI. Dr. Nour Khalil alerted. No outreach was proposed or sent.
Contactability
✓ Clear
Crisis flag
⚠ ESCALATE
AI judgment
NOT RUN — pre-agent gate
Override possible
No · hard stop
DEFERRED
req_6c0f9e8b · 22:14:02 UTCAn annual check-in arrived outside the configured send window. Operation ended · no queue · customer re-push required.
The message that didn't send is the product. SENT, HELD, ESCALATED or DEFERRED — the governed decision is the product. What most platforms show: what was sent. What REGENT shows: why — and what it stopped.
Every decision leaves a receipt
Decision record · sample · synthetic data
DecisionHELD
ReasonRecent human contact
AI proposalSEND
Policyuhc_reengagement.v3
Received10:41:08 UTC
Resolved10:41:11 UTC
Message sentNo
Not a black-box score. Not a campaign log. The evidence for why this person was — or was not — contacted. Structured. Linked to the dispatch. Integrity-verifiable on demand.
What REGENT guarantees
Audit trace per decision. Exact inputs, policy versions, model version, gates and outcome preserved.
Crisis flag = hard stop. Non-overridable at decision time. Named human alerted. No exception path.
Your rules run after the proposal. AI judgment is one input. Your governance policy is the authority.
Tamper-evident with governed redaction. The audit record is not simply immutable — it is integrity-verifiable.
Next step
See what REGENT would send — and what it would stop.
Run a Guided Eval →Start a design-partner pilotexautomatica.com/regent