I built an after-hours front desk that runs on the shop's own laptop, in the owner's voice, with no cloud AI
Description and use
A local voice receptionist for small trade shops answers after‑hours calls, judges urgency, and produces a summary card for the owner.
Problem
Small trade shops cannot staff a phone after hours and existing solutions are expensive or impersonal.
Outcome
The system achieved 95% accuracy on urgency judgment, 100% on emergency flagging, 98.75% on field extraction, and 90% on complete scenario accuracy across 20 scripted calls.
Workflow
- 1. Call comes in
- 2. Desk answers and speaks in the shop owner’s voice
- 3. Desk asks for problem, name, phone, address, requested time
- 4. Desk judges urgency against shop’s written rules
- 5. Desk records summary card at hang‑up
- 6. Card appears in owner’s morning review list
Evidence
The desk was not tuned against these scenarios first. They were run once each, as written.
Urgency judged right 19 of 20 (95%)
The desk answers out loud, asks what the problem is, takes a name, a callback number, an address, and a requested time
- Record date: 2026-10-14T03:41:09.656Z