AI Voice Booking Assistant

I designed a controlled AI booking conversation so a real service business can collect the right details, confirm pricing, and book appointments more reliably.

Functional Build10 stagesDocs published
10
Booking stages
6
Doc screenshots published
5+
Failure classes addressed
QA
Retest loop documented

Problem solved

Missed or unstructured calls left incomplete customer data, unclear service requests, pricing confusion, and unreliable appointment confirmation.

What I built

A 10-stage booking flow with one-question-at-a-time control, pricing confirmation checkpoints, address collection, appointment options, and a professional close.

Technologies used

Prompt engineering, conversational QA, business rules, ChatGPT; n8n / Airtable / Twilio in documented project scope.

Why this matters to an employer

You get hands-on conversational AI operations: define expected behavior, catch failure patterns, fix prompts/rules, retest, and document — the same loop production AI teams need.

Verified outcome

Documented 10-stage Functional Build with published sanitized docs. Addressed 5+ failure classes in testing (loops, repeated questions, order errors, pricing gaps, address handling). No call-volume metrics claimed.

Workflow diagram

Architecture diagram of booking stages — not a live telephony UI screenshot.

Architecture diagram of ten-stage voice booking flow

Architecture diagram — proprietary prompt text withheld.

Sanitized documentation evidence