DEV Community

Djaouad Frih
Djaouad Frih

Posted on

Case Study: Turning Missed Calls into Booked Leads With an AI Receptionist

The problem

A busy service business — clinic, salon, law firm — gets calls during the exact hours nobody's at the front desk. Every missed call is a lead that books somewhere else. Staff spend their day interrupting real work to answer the same five questions, then chase leads manually.

The fix

A 24/7 AI receptionist, trained on the business's own data. It doesn't pretend to be human. It does three things that directly move revenue:

  1. Never misses a lead — answers every inbound message in under a second, even at 2am
  2. Books, qualifies, routes — filters serious leads, books the slot into the calendar, routes to the right department
  3. Hands off with context — escalates to a human with an AI-drafted reply, so nothing is lost

How it ticks

The agent is grounded in a RAG knowledge base built from the business's own documents — services, prices, policies, FAQs. Everything it says is cited to real data, so it never invents prices or promises.

  • Multi-tenant: every business gets an isolated knowledge base
  • Streaming: answers start in <1 second; customers don't bounce
  • Native + web: one React Native codebase, plus an embeddable widget for any site or WhatsApp
  • Human handoff: the person gets a drafted reply with full context

The stack

Layer Choice
App (web) Next.js
App (mobile) React Native
Backend NestJS
Vector store PostgreSQL + pgvector
Realtime Socket.io
LLM Gemini

What "done" looks like

A business gains an always-on front desk that answers, books, and qualifies — and the human team only steps in where judgement actually matters. That's the difference between "AI demo" and "AI that does work."

See it live

This isn't a mockup — the agent is running in production on my own portfolio. Ask it about pricing, availability, or drop it a project brief:

→ Try the live agent: chat.djaouad.tech

Two more systems I run in production:

If you want this applied to your business: hire an AI developer — fixed price, milestone-based.

Top comments (0)