TUNED BY THE CALLS THEMSELVES
Richmond's iteration came from listening — to idle moments, to anonymous callers, to orders that almost did not make it. We worked with Ram to turn observed call behavior into product behavior.
THE CALLS TOLD US WHAT TO BUILD
Some of the most useful changes did not come from a feature request — they came from reviewing real Richmond calls and noticing where they stalled, where they dropped, and where an order was clearly intended but did not land cleanly.
Idle nudges
When a caller goes quiet, the AI generates a short, contextual nudge instead of letting the conversation stall — built from watching where real calls lost momentum.
Anonymous callers, handled on purpose
Calls from blocked or anonymous numbers get their own path: a clear message and a deliberate transfer rule, with sensitive details masked, so an anonymous caller is a decision, not an edge case.
Off-hours and smart transfers
Live-agent transfer stays available even outside business hours, and auto-transfer rules account for closing time and recent orders, so the call goes to a person at the moments that matter.
Order recovery when intent is clear
When a call ends with a clear order that did not land cleanly, a recovery process reconstructs the cart from the call's audio, so a detected order does not quietly disappear.
Conversations kept moving with contextual idle nudges
Anonymous callers routed deliberately, with details masked
Live-agent transfer available off-hours and around closing time
Orders reconstructed from audio when intent was detected
WANT THIS FOR YOUR MENU?
Book a meeting and we will walk through how AI Voice HQ would handle your menu, phone flow, and handoff rules.