Chat-PDF Website

Welcome to the Chat-PDF website project! This platform allows users to interact with their PDF documents through a chat interface. The project leverages various technologies to provide a seamless and efficient user experience.

Project Setup

To set up the project, follow these steps:

1. Copy the provided .env.example file to .env.

2. Generate API keys for the following services:

   • Clerk
   • PineconeDB
   • GeminiAPI
   • NeonDB
   • AWS

3. Set up your Next.js project by running the following commands:

# Install dependencies
npm install

npm run dev

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

Technologies Used

AWS S3 for PDF Storage

PDFs are securely stored on AWS S3, ensuring seamless access and reliability.

Pinecone DB for Metadata

Metadata extracted from the documents is intelligently organized using Pinecone DB. This enables efficient retrieval through vector search, making the search process faster and more accurate.

How VECTOR SEARCH Works

Vector search is a method based on representing documents as vectors in a high-dimensional space. Similar documents are close to each other in this space, enabling efficient and accurate search, even with a large number of documents.

NeonDB for User and Chat Data

The robust backend relies on NeonDB, managed via Drizzle ORM. This ensures seamless handling of user and chat data.

Gemini-AI API Integration

The power of conversation and text embedding is provided by the Gemini-AI API. A custom prompt is used to make conversations more natural, dynamic, and accurate.