How techwithtim/bdaiscraperagent works
Project type and purpose
• This is a full-stack AI travel planning application.
• Purpose: let users describe trips in natural language and receive flights, hotels, restaurant recommendations, and local insights in an interactive chat UI.
File structure
• Top‐level directories:
• frontend – the Streamlit UI, AI “assistant” modules, API client, local data and vector store
• backend – a Python service (“app.py”) for flight/hotel scraping + shared config/schemas
• frontend/restaurant_db – local ChromaDB vector store for restaurant embeddings
• Patterns:
• Clear separation between UI (Streamlit) & business logic (AI modules) & data access
• Backend acts as a microservice exposing endpoints for real-time flight & hotel search
• Config & build:
• requirements.txt + sample.env for dependencies/env
• No Dockerfiles, so local run via streamlit run and python app.py
Insights from the README
• Frontend uses Streamlit; AI calls go to Ollama/Claude; vector store is ChromaDB; scraping via BrightData; text search via DuckDuckGo API.
• Mentions two primary AI assistants (Travel Assistant, Research Assistant) plus a restaurant DB search.
• Backend primarily handles Google Flights & Hotels scraping modules.