How tanishka-birdie/gmail-rag works
Gmail RAG is a local inbox search-and-chat application. Its main workflow fetches Gmail messages, embeds and saves them locally, retrieves relevant messages for a question, and sends that context to Claude. The Gradio UI orchestrates both ingestion and multi-turn chat; a terminal query path is also documented. Edges follow the supplied source excerpts and README; local files and external services are shown as stores and integrations.
Main entry points: chatbot.py Gradio chatbot · query.py Vector retrieval · store_emails.py Embed and store · fetch_emails.py Gmail fetch
Components 10
Email ingestion
- Gmail fetch
gmail-RAG/fetch_emails.py - Embed and store
gmail-RAG/store_emails.py - Vectors and metadata
Search and answering
- Vector retrieval
gmail-RAG/query.py - Context and prompt
gmail-RAG/query.py
User interfaces
- Gradio chatbot
gmail-RAG/chatbot.py - Terminal query
gmail-RAG/query.py
External services
- Gmail inbox
- Claude API
Other components
- User
Connections 13
- User → Gradio chatbot: uses
README.md - Gradio chatbot → Gmail fetch: fetches via
gmail-RAG/chatbot.py - Gmail fetch → Gmail inbox: reads messages
gmail-RAG/fetch_emails.py - Gradio chatbot → Embed and store: stores via
gmail-RAG/chatbot.py - Embed and store → Gmail fetch: fetches emails
gmail-RAG/store_emails.py - Embed and store → Vectors and metadata: writes embeddings
gmail-RAG/store_emails.py - Gradio chatbot → Vector retrieval: retrieves via
gmail-RAG/chatbot.py - Vector retrieval → Vectors and metadata: loads vectors
gmail-RAG/query.py - Vector retrieval → Context and prompt: builds context
gmail-RAG/query.py - Context and prompt → Claude API: sends prompt
gmail-RAG/query.py - Gradio chatbot → Claude API: sends chat history
gmail-RAG/chatbot.py - User → Terminal query: asks questions
README.md - Terminal query → Vector retrieval: calls
gmail-RAG/query.py