WhatsApp AI Assistant
ClawBot (clawbot.ai) is a personal AI infrastructure that runs on your own hardware and integrates with 15+ messaging platforms including WhatsApp and Telegram. Inspired by ClawBot’s vision of ‘AI that lives on your terms’, we built a WhatsApp AI Assistant that allows users to chat with an intelligent AI directly from WhatsApp
The Problem
ClawBot addresses a fundamental problem — AI assistants are locked behind apps and websites. Users must open a browser, navigate to ChatGPT or similar tools, and context-switch from their daily workflow. This creates friction:
- People already spend hours on WhatsApp every day — but AI is not available there
- Opening a separate app or website breaks workflow and reduces AI adoption
- Most AI tools require internet browsers — not accessible to users who primarily use mobile messaging
- No conversation memory — every new session starts fresh, losing context of previous conversations
- AI is centralized on corporate servers — no privacy, no data sovereignty
The core problem: AI assistance should be available where people already are — on WhatsApp — not locked behind separate applications that require context switching.
Our Solution
Inspired by ClawBot’s messaging platform integration, we built a WhatsApp AI Assistant that brings AI directly into WhatsApp conversations:
- User sends any message on WhatsApp to the AI bot number
- Twilio receives the message and forwards it to the Next.js backend via webhook
- Groq LLM processes the message with full conversation history
- AI generates a helpful, contextual response
- Response is sent back to the user’s WhatsApp instantly
- Conversation history is maintained per user — AI remembers previous messages in the session
- Deployed on Render — available 24/7, no local computer required
The result: users get a powerful AI assistant accessible directly from WhatsApp — exactly as ClawBot envisions personal AI infrastructure that lives where users already communicate.
Solution Architecture
Complete Flow — Step by Step
| Step | Component | What happens |
| 1 | User (WhatsApp) | User sends message to Twilio sandbox number on WhatsApp |
| 2 | Twilio | Receives WhatsApp message, sends HTTP POST to /api/whatsapp webhook |
| 3 | Next.js API Route | Extracts From (phone number) and Body (message) from Twilio request |
| 4 | Conversation History | Map<phoneNumber, history[]> se user ki purani history fetch karo |
| 5 | Groq LLM | System prompt + history + user message bhejo, answer generate karo |
| 6 | History Update | New user message aur AI reply history mein save karo (last 20 messages) |
| 7 | Twilio API | client.messages.create() se WhatsApp pe reply bhejo |
| 8 | User (WhatsApp) | User ko WhatsApp pe AI ka jawab milta hai |
Key Components
- Twilio WhatsApp API — WhatsApp message receive aur send karne ka bridge
- Next.js /api/whatsapp route — webhook handler jo message process karta hai
- Groq LLM (Llama 3.3 70B) — AI brain jo actual answer generate karta hai
- In-memory Map — per-user conversation history storage
- Render deployment — 24/7 online, no local machine needed
ClawBot Alignment
- ClawBot: AI on messaging platforms → Our project: AI on WhatsApp via Twilio
- ClawBot: Conversation continuity across sessions → Our project: Per-user history Map
- ClawBot: Model-agnostic architecture → Our project: Groq (Llama 3.3 70B)
- ClawBot: Always available infrastructure → Our project: Render 24/7 deployment
Deliverables
- WhatsApp AI Assistant — fully functional, deployed on Render
- /api/whatsapp webhook route — Twilio integration with Next.js
- Per-user conversation history — Map-based memory, last 20 messages retained
- Groq LLM integration — fast, free, Llama 3.3 70B model
- Twilio WhatsApp Sandbox setup — message receive and send
- Render deployment — 24/7 availability without local server
- ngrok local development setup — for testing before deployment
- Environment-based configuration — all secrets in .env.local
Tech Stack
| Technology | Purpose | Why chosen |
| Twilio WhatsApp API | WhatsApp message receive + send | Industry standard, free sandbox, easy setup |
| Next.js 16 | Backend webhook handler | API routes, easy deployment on Render/Vercel |
| Groq API (Llama 3.3 70B) | AI response generation | Free tier, ultra-fast inference, high quality |
| LangChain Core | LLM message formatting | Standard AI message format (SystemMessage etc.) |
| Render | Cloud deployment | Free tier, always online, auto-deploy from GitHub |
| ngrok | Local development tunnel | Expose localhost to internet for Twilio webhook testing |
| TypeScript | Type-safe code | Prevents runtime errors in webhook handler |
| Node.js | Conversation history storage | Simple in-memory per-user history, no DB needed |
Business Impact
This solution brings AI directly into WhatsApp —
For Individuals
- AI assistance without opening any app or browser — just WhatsApp
- Works on any phone — even basic Android phones with WhatsApp
- Conversation memory — AI remembers context within a session
- Available 24/7 — no need to keep any computer running
For Businesses
- Customer support automation — businesses can deploy AI support on WhatsApp
- 2 billion+ WhatsApp users are potential users — no app download required
- Zero friction adoption — customers already use WhatsApp daily
- Cost effective — Groq free tier + Render free tier = zero running cost
Industries Impacted
- E-commerce — order tracking, product queries via WhatsApp
- Healthcare — appointment booking, symptom checking, medicine reminders
- Education — student queries, doubt solving, assignment help on WhatsApp
- Banking & Finance — balance queries, transaction alerts, financial advice
- Travel — booking assistance, itinerary help, real-time updates
