🤖 Echo.AI

Your Intelligent AI Companion for Real-time Web Search & Semantic RAG Document Q&A

Modern Chat UI · Google OAuth Authentication · LangChain Agentic Workflow · Tavily Web Search · PDF RAG Indexing

---

📖 What is Echo.AI?

Echo.AI is a robust, production-ready full-stack AI Assistant application. It delivers a fluid chat experience coupled with advanced cognitive capabilities:

• Web-search fallback capability using Tavily to supply contextually accurate, real-time data for general queries.
• Retrieval-Augmented Generation (RAG) pipeline allowing users to upload PDF documents and converse directly with them using vector search indexing on Pinecone.
• Intuitive UI controls such as chat pinning, session history management, and a clean, responsive layout built using Next.js 16 and Redux.

The application leverages a modular, feature-based architecture separating concerns clearly, enabling developers to scale components independently.

---

✨ Features

• 🔐 Google OAuth 2.0 Login — Seamlessly authenticate using Google credentials. The backend handles callback flows and issues JWTs in secure cookies.
• 💬 Dynamic AI Chatbot — Powered by Mistral AI (mistral-medium-latest) and orchestrated through LangChain's Tool-Calling Agent logic.
• 📑 Contextual PDF Q&A (RAG) — Upload documents (e.g. PDFs) within chat threads. The system extracts text content, splits it character-wise with overlaps, embeds it using Mistral Embeddings, and stores it inside Pinecone for instant lookup.
• 🌐 Real-time Web Search — Automated fallback search query triggering via Tavily AI Search when questions require recent or external internet data.
• 📌 Chat Management — Create, delete, and pin chats to keep important conversations accessible in the sidebar.
• 📁 Conversational Memory — Session-based chat history that maintains context across consecutive user queries.
• 📦 Containerized Workspace — Built with local Docker orchestration support for single-command environment setups.

---

🏗️ Tech Stack

Frontend (/client)

Technology	Purpose
Next.js 16	React framework & server-side rendering support
React 19	UI rendering engine
Redux Toolkit	Centralized state management (chats, auth, theme)
Tailwind CSS v4	Native utility-first stylesheet layouts
Motion (Framer)	High-performance CSS micro-animations & transitions
Spline 3D	Interactive 3D design components
Phosphor Icons	Premium icon design resources

Backend (/server)

Technology	Purpose
Node.js + Express 5	REST API router and controller hosting
MongoDB + Mongoose	Document Database and ODM modeling
LangChain JS	Agent orchestration, prompts, and vector chain utilities
Pinecone Client	Cloud-native vector search database handling
Passport.js	Social Google OAuth configuration
JSONWebToken	Secure, stateless authentication token handling
Multer	Multipart middleware for PDF file uploads
PDF-Parse	Server-side text extraction from documents

---

📂 Project Structure

Echo.AI/
├── client/                  # Next.js Frontend
│   ├── public/              # Static assets (favicons, manifest)
│   └── src/
│       ├── app/             # Routing directory (globals.css, layout, pages)
│       │   └── auth/        # Google Auth callback handler page
│       ├── components/      # UI components (ChatArea, ChatInput, Sidebar)
│       ├── features/        # Client business-logic features
│       │   ├── auth/        # Redux slices and hook wrappers for OAuth
│       │   └── chats/       # Chat operations, streaming Hooks, and APIs
│       └── store/           # Redux global store configurations
│
├── server/                  # Node.js + Express.js Backend
│   ├── uploads/             # Temporary folder for file uploads
│   └── src/
│       ├── app.js           # Server routing, CORS, and passport initializations
│       ├── config/          # Database connections and Env specifications
│       ├── controllers/     # Request handlers (auth, chat)
│       ├── dao/             # Data Access Objects (encapsulated database queries)
│       ├── middlewares/     # JWT verification and error captures
│       ├── models/          # MongoDB Mongoose schemas (User, Chat, Message)
│       ├── routes/          # API route definitions
│       ├── services/        # AI & LangChain agent orchestration
│       ├── tools/           # Tavily Web Search and Pinecone RAG handlers
│       └── utils/           # Shared helper utilities
│
└── docker-compose.yml       # Docker orchestration compose file

---

🔌 API Reference

🔐 Authentication — /api/auth

Method	Endpoint	Authorization	Description
GET	/google	Public	Redirects user to Google OAuth credentials screen
GET	/google/callback	Google Redirect	Processes Google OAuth profile data and issues JWT cookies
GET	/user	Bearer/Cookie	Resolves profile details of currently authenticated user
POST	/logout	Bearer/Cookie	Clears auth tokens and logs out user session

💬 Chat & RAG — /api/chat

Method	Endpoint	Authorization	Description
GET	/	Bearer/Cookie	Fetches all chat sessions for sidebar lists
POST	/	Bearer/Cookie	Sends user query. Supports multipart forms with an optional file attachment pdf
GET	/:chatId	Bearer/Cookie	Retrieves complete conversational message history of a chat
DELETE	/:chatId	Bearer/Cookie	Deletes the chat session and clears related stored messages
PATCH	/:chatId/pin	Bearer/Cookie	Toggles the pinned status on the specified chat session

---

⚙️ Local Setup

Prerequisites

• Node.js (version 20+)
• MongoDB Atlas database connection URL
• Google Cloud developer credentials (for Google OAuth client ID and Secret)
• Mistral AI developer API Key
• Tavily search API Key
• Pinecone DB index named kodr-rag with 1024 dimensions (compatible with mistral-embed)

---

Method 1: Running Manually (Locally)

1. Clone the repository

git clone https://github.com/shobhit2603/Echo.AI.git
cd Echo.AI

2. Set Up the Backend Server

cd server
npm install

Create a .env file inside /server:

PORT=5000
NODE_ENV=development
MONGO_URI=your_mongodb_connection_string
JWT_SECRET=your_jwt_signing_key_secret
CLIENT_URL=http://localhost:3000

# Google OAuth Credentials
GOOGLE_CLIENT_ID=your_google_client_id.apps.googleusercontent.com
GOOGLE_CLIENT_SECRET=your_google_client_secret
GOOGLE_CALLBACK_URL=/api/auth/google/callback

# AI Keys
MISTRAL_API_KEY=your_mistral_api_key
TAVILY_API_KEY=your_tavily_api_key
PINECONE_API_KEY=your_pinecone_api_key

Run the backend in development mode:

npm run dev

The server will boot up and listen on http://localhost:5000.

3. Set Up the Client Frontend

cd ../client
npm install

Create a .env file inside /client:

NEXT_PUBLIC_SERVER_URL=http://localhost:5000

Start the Next.js development server:

npm run dev

Open http://localhost:3000 in your browser.

---

Method 2: Running with Docker-Compose

To spin up both frontend and backend instantly with live-reloading (hot reloading) capabilities:

1. Ensure you have .env files configured

Make sure you have created /server/.env and /client/.env files containing the relevant keys (as defined in the manual setup section).

2. Launch Docker Compose

In the root directory of the project, execute:

docker-compose up --build

This command automatically reads the docker-compose.yml file, builds the server and client images from their respective dockerfiles, installs dependencies, mounts development directories with node_modules caches, and starts the services.

• Frontend client is exposed on: http://localhost:3000
• Backend server is exposed on: http://localhost:5000

To stop the containers, run:

docker-compose down

---

🌐 Deployment

Service	Host Provider	Configuration
Frontend Client	Vercel	Connect your Git repository, set the root directory to client, configure NEXT_PUBLIC_SERVER_URL in environment variables.
Backend Server	Render	Create a Web Service, specify base directory as server, and configure environment variables.
Database	MongoDB Atlas	Managed MongoDB Cloud Database
Vector DB	Pinecone	Cloud Vector DB storing 1024-dimension embeddings

Note

Ensure that you white-list production URLs in your Google Developer console redirects (/api/auth/google/callback) and that the CORS allowed origin on Render aligns with your deployed Next.js address.

---

📄 License

Distributed under the MIT License. See LICENSE for more details.

---

Built with ❤️ by Shobhit