Re-engagement outreach

uhc_reengagement.v3 · active · effective 28 Jul 2026 · channels: whatsapp

DEMO · SYNTHETIC DATAThis console is a public product tour served from synthetic fixtures. There is no login, no tenant credential in the page, and no connection to a Regent backend. A real tenant console requires a server-side session boundary before a browser may reach tenant-scoped APIs.

Pre-judgment controls · run before the judgment engine

ContactabilityActive channel required
A subject with no active, reachable channel never reaches the judgment engine.
Crisis flagNon-overridable hard stop
A crisis flag present in the accepted snapshot resolves escalation before the judgment engine is invoked. There is no operator override at decision time.
Global opt-outNon-overridable hard stop
A global opt-out on record stops the operation before any proposal is generated.

Post-judgment controls · run after the AI proposal, before any dispatch

Proposal-specific consentScope must cover the proposed intent
Consent is evaluated against the specific proposal, not against a blanket flag. Tenant-asserted evidence, provenance-labelled.
Recent human contactHold within 48 hours
If a human on the care team spoke to the subject inside 48 hours, AI outreach is held. Terminal.
Frequency cap3 per rolling 7 days
Counted by Regent from its own dispatch history, not from tenant-asserted counters.
Send window08:00–21:00 subject-local
Opt-in window. Pushes received outside it resolve DEFER_REPUSH_REQUIRED — the operation ends and the customer re-pushes. Nothing is queued.
Content policyPermitted topic classes only
Forbidden-content linting runs on the proposal before dispatch is authorized.

Escalation

Named targetDr. Nour Khalil · on-call clinical lead

Acknowledgment SLA15 minutes

FallbackLadder to a secondary target, then the named operator alarm

Decisions governed by this version

Disabled in the demo. This action belongs to a signed backend sprint. Policy mutation is a Sprint 4 capability with its own signed contract.