Hey everyone,
I wanted to share a project I’ve been developing called Chat with Bella — a real-time travel assistant that runs entirely on WhatsApp.
It uses:
-
GPT-4o for conversational responses (multi-language, context-aware)
-
Whisper for transcribing voice notes
-
Redis for chat history and daily message limits
-
Slack logging for monitoring performance
-
Flask API hosted on Ubuntu for webhook handling
The system manages real user interactions — answering travel questions, sending Google Maps links, and providing restaurant tips.
Here’s a simplified flow diagram:
User → WhatsApp → Meta Webhook → Flask → GPT-4o → Reply → WhatsApp
I’m currently exploring ways to improve:
-
Context compression for long conversations
-
Multi-channel expansion (Telegram, iMessage, etc.)
Would love to hear how others handle session context + rate limiting in production chatbots.