Skip to content

Trust Center

An AI front desk is only useful if you can trust it.

The biggest fear owners share with us is the AI saying something wrong. Here's exactly how we prevent it, and what we're honest about still building.

The AI proposes. Your rules approve.

Every booking, price, deposit and cancellation passes through a deterministic policy engine built from your settings. The language model can't book a service you don't offer, at a price you didn't set, in a slot that's taken.

Confidence-based human handoff

Low confidence, an upset caller, a refund, a complaint, a medical or safety question: the AI stops and hands off, by warm transfer or callback, with a summary so nobody repeats themselves.

Always disclosed

Every conversation starts by saying it's an AI assistant. We don't pretend to be human. Transparency rules like the EU AI Act's chatbot obligations make this the law in more places, and we think it's simply right.

Every action on the record

Transcripts, recordings (with consent notice), and a log of every action the AI took, available in your dashboard and exportable.

Your data stays yours

Encrypted in transit and at rest. We don't sell data or use your clients' conversations to market to them. One-click export, and deletion on request.

Tested before every release

An evaluation suite of real-world scenarios (accents, interruptions, rescheduling, emergencies, trick questions) must pass before any agent update ships.

Compliance status

We label what's done and what isn't. We'll never claim a certification we don't hold.

AI disclosure at conversation start

All channels

In place

TCPA consent capture & SMS opt-out (STOP)

Consent text versioned and stored with timestamp

In place

A2P 10DLC registered messaging

Required for US business texting

In progress

Call-recording consent notices

Played at call start; state rules respected

In place

SOC 2 Type I

Roadmap target after pilot program

Planned

HIPAA safeguards & BAA

Required before any healthcare pilots

In progress

Security questions or a vendor questionnaire? Email security@agilexy.com.

See the guardrails in action.

Call the demo line and ask for a refund, or tell it you smell gas. Then judge for yourself.

Call AI demoStart free pilot