DocuMancer Workflow

When user sends a query request to FastAPI backend, the backend will call the RAG pipeline along with query and generates response back to user.

🧠 Under the Hood: The DocuMancer Workflow

Here’s how it flows:

1. You ask a question in the chatbot.
2. The backend:
   • Loads your .md files.
   • Breaks them into small chunks (for better LLM understanding).
   • Stores them in a vector database (FAISS).
3. When a query comes in:
   • It retrieves the top relevant document chunks.
   • Feeds them to the LLM with your question.
   • LLM generates an accurate, context-based response.
4. You get the answer + source files + token cost.