WhaBot AI - WhatsApp Customer Service

``` WhaBot AI is a complete WhatsApp customer service SaaS platform that uses GPT-4o, Claude, and Gemini to automatically respond to customer inquiries. Connect your WhatsApp Business number and let AI handle 80% of support — with smart handoff to human agents when needed.

Built on the official WhatsApp Cloud API (Meta Graph API v21.0) — no unofficial libraries, 100% compliant.

Multi-Model AI Engine: GPT-4o, Claude Sonnet, Gemini Pro — switch per organization. Knowledge-augmented responses with confidence scoring and automatic human handoff when AI is unsure.

Knowledge Base: Create unlimited articles organized by category. Keyword-based search with TF-IDF scoring. AI automatically references relevant articles in responses.

Conversation Flow Builder: Keyword triggers, first-message triggers, menu triggers. Actions include send text, buttons, lists, AI response, assign agent, add tag, and close conversation.

Live Agent Handoff: Confidence threshold-based escalation, toggle AI on/off per conversation, agent assignment, quick replies for agents.

CRM & Contact Management: Auto-created contacts from WhatsApp messages, tags, notes, conversation history, search and filter.

Analytics Dashboard: Message volume charts, AI resolution rate, response time, open vs resolved conversations, credit usage tracking.

Stripe Billing: 3 subscription tiers (Starter $29, Pro $79, Enterprise $249), credit-based AI usage, credit top-up via Checkout.

Multi-Tenant: Organizations with team members, role-based access (Owner, Admin, Agent), per-org settings. ```

Features

``` - Multi-Model AI: GPT-4o, Claude Sonnet, Gemini Pro — switch per organization - Official WhatsApp Cloud API (Meta Graph API v21.0) - Send text, images, documents, audio, video, interactive buttons, list messages - Knowledge base with category organization and TF-IDF search - Conversation flow builder with keyword/menu/first-message triggers - Live agent handoff with confidence threshold - CRM with auto-created contacts, tags, notes, conversation history - Analytics dashboard with message volume, resolution rate, response time charts - Stripe billing with 3 subscription tiers and credit system - Multi-tenant architecture with role-based access (Owner, Admin, Agent) - Template messages for proactive WhatsApp outreach - Webhook signature verification (HMAC-SHA256) - Quick replies for agents (/greeting, /thanks, /resolved) - Dark mode responsive UI - MIT License ```

Requirements

``` - Node.js 18 or higher - npm or yarn - No database server needed (SQLite included by default) - WhatsApp Business API credentials from Meta (for production) - At least one AI API key (OpenAI, Anthropic, or Google) - Stripe account for billing features - Optional: PostgreSQL for production deployment ```

Instructions

``` 1. Extract the ZIP file 2. Run: npm install 3. Run: npx prisma db push 4. Run: npx tsx prisma/seed.ts (loads demo data with articles, flows, templates) 5. Run: npm run dev 6. Open http://localhost:3000 7. Login with admin@whabot.ai / admin123 (999 credits) 8. Dashboard shows overview with stats and charts 9. Go to Knowledge Base → Add articles for AI to reference 10. Go to Flows → Create conversation triggers and actions 11. Go to Settings → Configure WhatsApp number and AI model 12. For WhatsApp integration: Set webhook URL to /api/webhook/whatsapp in Meta dashboard

Agent login: demo@whabot.ai / demo123 (50 credits) For production: Set DATABASE_URL to PostgreSQL, configure Stripe and WhatsApp API keys. ```