Regent
Governed engagement engine.
Most AI engagement tools tell you what was sent. Regent tells you why it was right to send. Your rulebook runs in a fixed order: hard stops first — contactability, crisis state, global opt-out — and a hard stop resolves the operation before your AI is ever invoked. Only a case that clears them reaches the judgment engine. Whatever it then proposes runs through the proposal-specific controls — consent, content policy, quiet hours, frequency — before anything reaches a customer. If the rules say no, the message never goes.
Every decision is recorded in a trace that is tamper-evident with governed redaction: the exact inputs, policy versions, model version, gates and recorded outcome are preserved and integrity-verifiable on demand. Built for healthcare and fintech organizations in MENA operating under WhatsApp Business API at scale.
Hard stops first
Contactability, crisis flag and global opt-out are evaluated before judgment. A crisis flag present in the accepted context resolves an escalation and the AI is never invoked — there is no override at decision time.
Then AI proposes
Only a case that cleared the hard stops reaches your judgment engine. It recommends whether and how to reach out. Regent doesn't replace it — it governs what happens next.
Then your rules govern
Proposal-specific controls — consent for this intent, content policy, quiet hours, frequency — run independently after the proposal and before any send.
Every outcome is on the record
A tamper-evident trace records every check, every gate result, and the final decision. Integrity-verifiable on demand. Every decision leaves a receipt.
Status: in build and pilot preparation. UniversalHealthCare.ai is the first proving tenant; trial go-live has not happened yet.