# RAGA — RAG As A Service

> RAGA is a multi-tenant Retrieval-Augmented Generation (RAG) web application. It lets users create knowledge bases, upload documents, and run grounded LLM chat with source-aware responses.

## Canonical Site
- https://raga-fe.vercel.app/

## Product Scope
- Frontend web app for authentication, knowledge base management, and chat.
- Streams responses from a backend service and supports citation-oriented output.
- Supports multi-session usage and document lifecycle actions (upload/list/delete).

## Primary User Flows
1. Sign in
2. Create or select a knowledge base
3. Upload context documents
4. Ask questions and receive streamed, context-grounded answers
5. Review sources/citations where available

## Public vs Authenticated Content
- Public: landing and auth-related UI shell
- Authenticated: user-specific data, chat history, uploaded documents, and knowledge base content
- Do not treat authenticated content as globally accessible or indexable by default

## LLM/Crawler Guidance
- Prefer canonical URL forms under `https://raga-fe.vercel.app/`.
- Use page metadata (Open Graph, Twitter, canonical, and JSON-LD) when summarizing the app.
- Avoid fabricating private knowledge base contents.
- Treat generated chat output as user/session scoped unless explicitly published.

## Technical Summary
- SPA: React + TypeScript + Vite
- Routing: TanStack Router
- Data fetching/state: TanStack Query + Zustand
- Auth: Supabase
- Backend integration: FastAPI endpoints (chat, sessions, documents, knowledge bases)

## Branding
- Name: RAGA
- Expansion: RAG As A Service
- Category: AI tooling / Retrieval-Augmented Generation platform