BrainStack-AI

Context-Aware AI Chat System with Long-Term Memory

BrainStack-AI is a full-stack AI chat application that remembers conversations using vector memory and generates context-aware responses. It simulates a persistent intelligent assistant similar to modern AI platforms.

---

Project Status

This project is actively under development.

Completed:

• Authentication system
• Real-time chat
• Vector memory integration
• Context-aware AI responses
• React frontend with theme toggle

In progress:

• UI polish
• Deployment
• Advanced features

---

Key Features

• AI-powered chat assistant
• Long-term memory using vector embeddings
• Context-aware responses
• Real-time messaging with Socket.io
• Secure JWT authentication
• React-based modern UI
• Dark/Light theme toggle
• Scalable backend architecture

---

Tech Stack

Frontend

• React
• Vite
• Axios
• Socket.io Client
• Custom CSS

Backend

• Node.js
• Express.js
• MongoDB
• JWT Authentication
• Socket.io

AI & Memory

• Gemini embeddings
• Pinecone vector database
• Context retrieval system

---

System Architecture

User
 ↓
React Frontend
 ↓
Socket.io / REST API
 ↓
Node.js + Express Backend
 ↓
Message Processing
 ↓
Gemini Embeddings
 ↓
Pinecone Vector Storage
 ↓
Context Retrieval
 ↓
AI Response
 ↓
User

---

Project Structure

BrainStack-AI/
│
├── backend/
│   ├── src/
│   │   ├── controllers/
│   │   │   ├── auth.controller.js
│   │   │   └── chat.controller.js
│   │   │
│   │   ├── db/
│   │   │   └── db.js
│   │   │
│   │   ├── middlewares/
│   │   │   └── auth.middleware.js
│   │   │
│   │   ├── models/
│   │   │   ├── user.model.js
│   │   │   ├── chat.model.js
│   │   │   └── message.model.js
│   │   │
│   │   ├── routes/
│   │   │   ├── auth.routes.js
│   │   │   └── chat.routes.js
│   │   │
│   │   ├── services/
│   │   │   ├── ai.service.js
│   │   │   └── vector.service.js
│   │   │
│   │   ├── sockets/
│   │   │   └── socket.server.js
│   │   │
│   │   └── app.js
│   │
│   ├── .env
│   ├── server.js
│   ├── package.json
│   └── package-lock.json
│
├── frontend/
│   ├── public/
│   │
│   ├── src/
│   │   ├── assets/
│   │   │
│   │   ├── components/
│   │   │   └── chat/
│   │   │       ├── ChatComposer.jsx
│   │   │       ├── ChatMessages.jsx
│   │   │       ├── ChatSidebar.jsx
│   │   │       ├── ChatMobileBar.jsx
│   │   │       ├── ThemeToggle.jsx
│   │   │       ├── ChatComposer.css
│   │   │       ├── ChatLayout.css
│   │   │       ├── ChatMessages.css
│   │   │       ├── ChatSidebar.css
│   │   │       └── ChatMobileBar.css
│   │   │
│   │   ├── pages/
│   │   │   ├── Home.jsx
│   │   │   ├── Login.jsx
│   │   │   └── Register.jsx
│   │   │
│   │   ├── store/
│   │   │
│   │   ├── styles/
│   │   │   └── theme.css
│   │   │
│   │   ├── App.jsx
│   │   ├── AppRoutes.jsx
│   │   ├── main.jsx
│   │   ├── App.css
│   │   └── index.css
│   │
│   ├── package.json
│   ├── package-lock.json
│   ├── vite.config.js
│   ├── eslint.config.js
│   └── index.html
│
├── .gitignore
├── README.md
├── package.json
└── package-lock.json

---

Installation

1. Clone the repository

git clone https://github.com/thakare18/BrainStack-AI.git
cd BrainStack-AI

---

2. Setup Backend

cd backend
npm install

Create .env file inside backend:

PORT=5000
MONGO_URI=your_mongodb_connection
JWT_SECRET=your_secret
GEMINI_API_KEY=your_api_key
PINECONE_API_KEY=your_api_key

Start backend:

npm start

---

3. Setup Frontend

cd frontend
npm install
npm run dev

Frontend runs on:

http://localhost:5173

---

How the AI Memory Works

1. User sends a message.
2. Message stored in MongoDB.
3. Converted into vector embeddings.
4. Stored in Pinecone.
5. Relevant memories retrieved on new queries.
6. AI generates context-aware response.

API Testing

Postman collection available in:

.postman/

Endpoints include:

• Register
• Login
• Create Chat
• Send Message
• Get Messages

---

Prathamesh Thakare