0504.1.mp4

WhatsApp AI Automation

WhatsApp AI Automation is a portfolio-grade customer service assistant for WhatsApp. It combines a Baileys WhatsApp client, an Express API, a Prisma/PostgreSQL data layer, Groq LLM responses, and a Next.js dashboard shell.

Status: active MVP. The backend message loop, AI pipeline, Prisma schema, seed data, and dashboard foundation are implemented. Some admin dashboard features are still planned.

Features

• WhatsApp message ingestion with Baileys QR login
• AI responses through Groq Llama models
• RAG-lite FAQ retrieval from PostgreSQL
• Human handoff detection with configurable keywords
• Conversation and message persistence with Prisma
• Multi-business data model foundation
• Next.js dashboard foundation with Clerk authentication

Tech Stack

Layer	Technology
API	Node.js, Express, TypeScript
WhatsApp	Baileys
AI	Groq SDK, Llama 3.1
Database	PostgreSQL, Supabase, Prisma
Web	Next.js 14, Tailwind CSS, Clerk
Quality	Node test runner, TypeScript, GitHub Actions

Repository Structure

apps/
  api/        Express API, WhatsApp client, AI pipeline, Prisma schema
  web/        Next.js dashboard
packages/
  shared-types/
docs/         Architecture notes, roadmap, interview explanations

Prerequisites

• Node.js 18 or newer
• PostgreSQL database, local or Supabase
• Groq API key
• WhatsApp account for QR pairing

Quick Start

1. Install dependencies:

npm install

2. Copy environment variables:

cp .env.example .env

3. Fill the required values:

DATABASE_URL="postgresql://postgres:password@localhost:5432/whatsapp_ai"
GROQ_API_KEY="your_groq_key_here"

4. Generate Prisma Client:

npm run prisma:generate

5. Apply your database schema and seed demo data:

npm run seed

6. Start the API:

npm run dev --workspace=@wa-automation/api

7. Scan the QR code shown in the terminal to connect WhatsApp.

8. Start the web dashboard in another terminal:

npm run dev --workspace=@wa-automation/web

Demo Data

The seed script creates:

• A default demo business with ID default
• AI settings for Groq
• FAQ examples for opening hours, location, and menu recommendations
• A sample conversation with inbound and outbound messages

This matches the current API code path, which uses businessId = "default" for incoming WhatsApp messages.

Useful Commands

npm run typecheck
npm test
npm run build
npm run prisma:generate
npm run seed

Security Notes

Do not commit .env, WhatsApp session folders, QR screenshots, phone numbers, or real customer conversations.

Baileys stores WhatsApp credentials locally. This repo ignores .wa-auth/, auth/, and auth_info_baileys/. If a session file was committed before, remove it from Git tracking and reconnect WhatsApp to generate a fresh session.

Roadmap

• Admin API endpoints for conversation and FAQ management
• Dashboard integration with live API data
• FAQ CRUD UI
• Analytics for message volume, handoff rate, and FAQ usage
• Queue-based message processing for production workloads
• Optional pgvector retrieval for larger knowledge bases
• Deployment guide for Railway, Vercel, and Supabase

License

MIT