AI Travel Chatbot Prototype

I’m building an MVP of an AI-powered travel itinerary planner that works both on our website and over WhatsApp. The front-end is already live and cloud resources are provisioned; what I need now is the core Python back-end that ties everything together.

Core scope
• Python (Flask or Django) service that exposes a chat endpoint for the web client and connects to the WhatsApp Business API.
• Retrieval-Augmented Generation pipeline: scrape chosen travel sites and blend that content with a small internal database you’ll help me model, then embed it for fast vector search before each LLM call.
• Named-Entity Recognition on every assistant response so the site can turn places, dates and activities into simple “travel cards” (see mindtrip.ai for the interaction style).
• Persistent storage of chat history and basic user profiles so conversations can be resumed seamlessly.

Minimum viable output
1. Running prototype deployed to my cloud account.
2. Clear setup scripts / Dockerfile so I can reproduce locally.
3. Brief schema diagram and code comments showing where to plug in extra data sources or add booking flows later.
4. The output should be parsed and travel cards should be created based on the output - Very important feature

The chatbot should be able to:
– answer destination queries,
– surface travel tips,
– trigger a placeholder “book now” action.
- Create travel cards on the website. Each travel card would represent the location/ experience or itinerary.

Keep it functional and lightweight; no fancy UI work is needed on your side. If you have questions about the data sources or preferred libraries (LangChain, Haystack, spaCy, etc.), just ask and I’ll clarify.