All work
Customer operations

Yelp Booking Agent

An AI booking agent that moves Yelp leads from first response through qualification, live availability, and confirmed CRM jobs.

My role
Conversation architecture, booking state, CRM integration, testing tools
System
React, TypeScript, Express, OpenAI, Yelp, ServiceTitan, Firebase
The shared inbox shows lead context, AI activity, staff messages, and booking state.
01

Lead context

Yelp leads often arrive with partial information. The system has to understand the requested service, collect location and contact details, apply business rules, and move quickly enough that the lead does not go cold.

The inbox keeps the original request, conversation, booking state, and staff activity in one place, so every lead has a visible path to resolution.

02

Booking path

  1. 01Lead message
  2. 02Qualification
  3. 03Customer details
  4. 04Live availability
  5. 05Hold and claim
  6. 06CRM job
03

Lab mode

Lab mode lets operators test realistic lead scenarios against the same business rules and booking logic used in production, making response quality inspectable before changes go live.

Lab mode used to inspect how the agent responds to a service request.
04

Business rules

Service ZIP codes, business accounts, response signatures, automation mode, and tenant behavior live in explicit settings, giving operators direct control over how the agent works.

Business settings for service areas and tenant-level behavior.
05

Operational closure

Exception queue

Unsupported work, incomplete details, policy questions, and CRM failures become visible staff tasks with the conversation context attached.

Confirmed jobs

Availability is refreshed and booking state is persisted before the customer receives a confirmation.