AI & Voice Disclosure

APEX VoiceOrder is designed so that, when it answers a restaurant's phone, callers interact with an automated AI voice assistant, not a live person. Restaurants configure how this is disclosed on the call, and a human on the restaurant's team can be reached through escalation.

The AI voice model handles the conversation: understanding the caller, natural phrasing, intent extraction, clarification, and recovery. It does not invent the facts that matter.

Prices, taxes, fees, discounts, totals, item and modifier availability, payment status, order acceptance, and completion status are controlled by deterministic Apex Intelligence Analytics services — not generated by the model. If a fact isn't confirmed by those services, the order does not proceed; the assistant clarifies or hands off to a person.

When a call falls outside what the assistant can complete confidently, it transfers the call to the restaurant's staff with context. The system is designed to escalate rather than fail silently.

To confirm orders and improve quality, calls may be recorded and reviewed, subject to the restaurant's configuration and applicable law. Recording practices and any required notices are set up per restaurant during onboarding.

We describe capabilities we've set out to deliver and validate — including support for U.S., Indian, and Pakistani English — without guaranteeing perfect recognition of every accent, dialect, or request. Where the system isn't confident, it clarifies or hands off. We don't publish accuracy percentages we haven't earned.

At launch, ordering runs on pay-at-pickup or an optional SMS payment link. The voice assistant does not collect card details by voice. Secure in-call payment collection is a planned future capability, not a current one.