DocuSense AI

Overview

This repository contains a React + Vite frontend and an Express-based Node server for a document intelligence application.

The project uses:

• React 19 + Vite
• Express server bundled with esbuild
• Google Gemini AI via @google/genai
• Tailwind CSS for styling
• Vitest for tests

Prerequisites

• Node.js 20 or later
• npm

Setup

1. Install dependencies:

   npm install

2. Create a .env file in the project root with the following values:

   GEMINI_API_KEY=your_google_gemini_api_key
   JWT_SECRET=your_jwt_secret

   • GEMINI_API_KEY is required for AI model requests.
   • JWT_SECRET is used by the server for token generation and validation.

Development

Run the app in development mode:

npm run dev

This starts the application via tsx server.ts and serves the frontend through Vite.

Build

To build the frontend and bundle the server for production:

npm run build

What this does:

• vite build builds the React frontend
• esbuild server.ts --bundle --platform=node --format=cjs --packages=external --sourcemap --outfile=dist/server.cjs bundles the Express server into dist/server.cjs

Start

After building, start the production server with:

npm run start

The server listens on port 3000 by default.

Clean

Remove build artifacts:

npm run clean

Test

Run the test suite:

npm run test

Notes

• The server uses an in-memory database, so data does not persist across restarts.
• File upload size limits are configured in server.ts with JSON and URL-encoded payload limits of 50mb.
• If you need to customize the port, update PORT in server.ts.

Project Info

DocuSense AI is designed for enterprise document analysis and decision support. It combines:

• A React frontend for dashboarding, document upload, and knowledge graph visualization
• An Express backend for API handling, AI requests, authentication, and document parsing
• Google Gemini AI integration for classification, summarization, and compliance insights
• In-memory state for user sessions, documents, conversations, audit logs, and shared reports

This build is intended as a modern proof-of-concept stack with fast local development and easy production bundling.

Next Uses

1. Extend the AI flows:

   • add smarter prompt templates
   • support additional Gemini models or fallback logic
   • persist results to a database instead of memory

2. Add production readiness:

   • implement real authentication and user management
   • add logging, metrics, and error tracking
   • secure env handling and secrets management

3. Improve UX:

   • add file upload previews and document history
   • enhance the knowledge graph and analysis viewer
   • support PDF, image OCR, and advanced document types

4. Deploy:

   • containerize with Docker
   • host on a cloud platform (AWS, GCP, Azure, or Vercel for frontend)
   • connect to a persistent database and secure API gateway