PDF AI Assistant: A Full Stack SaSS RAG (Retrieval Augmented Generation) Application

Introduction

The SaSS application is based on the youtube video with the secondary development.
A modern full-stack web application that allows users to upload PDF files, perform hybrid search on the contents, and engage in AI-powered chat conversations powered by Google Genimi.

Highlights

• MongoDB Atlas for storage, vector search, keyword search and customized hybrid search.
• Backend APIs for file uploading,context retrieval deployed by CloudRun and AWS Lambda.
• Google Genimi API for chat generation
• Purely chat mode available to disable reading PDF
• Jina AI for embedding and retrieving reranking
• UI optimizations for model devices

Preview

www.chat-pdf-ai.com

Overall Architecture

Backend API Github: https://github.com/Nelsonlin0321/chat-pdf-ai-assistant-fastapi

Backend:

• The FastAPI server acts as the main backend API, handling complex logic.
• It uses the Jina framework for embedding text data and reranking for retrieval data.
• MongoDB act the the backend database and perform vector similarity and keyword search operations for ingested vector and text.
• The custom "Read & Chunk" component to preprocessing PDF files, splitting pages to sentence and them merge them into chunks for embedding and indexing.
• The "api/hybrid_search" route handle hybrid search queries, combining traditional text search (BM25) with vector similarity search with Jina AI reranking.

Frontend:

• The frontend is built with the Next.js React framework, providing server-side rendering and static site generation capabilities.
• It integrates with third-party services like Stripe for payment processing, Clerk for user authentication (sign-in, sign-up), and an AI SDK for chat functionality (api/ai/chat).
• The frontend communicates with the backend API for CRUD operations, managing data related to the application's core functionality.
• Vercel is used for hosting and deploying the Next.js frontend application.

Data Flow:

• PDF files are ingested through the backend API (api/ingest_file) and processed (read, chunked, embedded) before storing in the vector database.
• User queries (questions) are embedded using the Jina API, and vector similarity search is performed against the ingested vector database to retrieve relevant chunks or documents.
• The retrieved chunks are re-ranked using the Rerank Jina API and other algorithms (BM25) to produce the final search results.
• The frontend interacts with the backend API to fetch and display these search results, as well as to handle other functionality like authentication, payments, and AI-powered chat.

Run Locally

npm run install
npm run dev

Open http://localhost:3000 with your browser to see the result.

env

# Clerk
NEXT_PUBLIC_CLERK_PUBLISHABLE_KEY
CLERK_SECRET_KEY=
NEXT_PUBLIC_CLERK_SIGN_IN_URL=/sign-in
NEXT_PUBLIC_CLERK_SIGN_UP_URL=/sign-up
NEXT_PUBLIC_CLERK_AFTER_SIGN_IN_URL=/
NEXT_PUBLIC_CLERK_AFTER_SIGN_UP_URL=/

# DATABASE_URL
DIRECT_DATABASE_URL=
DATABASE_URL=

NEXT_PUBLIC_S3_ROOT_DIR=chatpdf

NEXT_PUBLIC_LAMBDA_BACKEND_URL=
NEXT_PUBLIC_CLOUDRUN_BACKEND_URL=

GOOGLE_API_KEY=

STRIPE_API_KEY=

STRIPE_WEBHOOK_SIGNING_SECRET=

NEXT_BASE_URL="http://localhost:3000"

SENTRY_AUTH_TOKEN=