🔮 LLM Workflow Agent — Hindi-English AI Workflow Assistant

Built for WCHL Hackathon 2025 | Qualified for National Level

Automates everyday tasks via Hindi-English mixed prompts using LLM and APIs.
Handles reminders, emails, todos, weather queries, Google Calendar scheduling with Meet links.

---

🧠 Problem Statement

Issue	Description
Task Fragmentation	Users manage reminders, emails, todos, and calendar events separately.
Language Barrier	Existing assistants poorly support mixed Hindi-English commands.
Manual Effort	Daily workflow tasks require repetitive manual intervention.
Objective	Build an AI agent that understands mixed-language commands and automates common workflow tasks efficiently.

---

💡 Our Solution — LLM Workflow Agent Platform

Component	Description
LLM Engine	LangChain + Google Gemini integration to parse and understand Hindi-English prompts.
Task Agents	Modular task handlers for emails, reminders, todos, weather, and calendar events.
Scheduler	Cron-like task scheduler for reminders and calendar events.
Backend API	Node.js + Express.js API exposing endpoints for frontend and external integrations.
Frontend Dashboard	React + Vite + Tailwind interface for user interaction, chat fallback, and task overview.
Realtime Updates	Socket.IO for live reminder alerts and task notifications.

---

🧩 Core Features

Feature	Description
🗣️ Mixed-Language Understanding	Supports Hindi-English commands for natural task input.
🔔 Reminders & Alarms	Schedule tasks with live alerts via Socket.IO.
📧 Smart Email Automation	Compose and send emails automatically using Nodemailer.
☀️ Weather Queries	Fetch live weather info for any city.
✅ Todo Management	Create, view, and manage todos seamlessly.
🗓️ Google Calendar Integration	Schedule events with auto-generated Meet links.
💬 LLM Chat Fallback	Handles general conversation if no specific task detected.

---

⚙️ Tech Stack

Layer	Technology Used
Backend	Node.js, Express.js, JWT, Session Middleware
AI / LLM	LangChain + Google Gemini API
Database	MongoDB (Mongoose)
Email	Nodemailer
Calendar & Meet	Google Calendar API
Frontend	React, Vite, Tailwind CSS
Realtime	Socket.IO

---

📂 Project Structure

nl_task_automator/
├── backend/
│   ├── agents/         # Task logic modules (email, calendar, reminders, etc.)
│   ├── controllers/    # Express route handlers
│   ├── llm/            # Gemini API integration and prompt handling
│   ├── middlewares/    # Authentication and error handlers
│   ├── models/         # Mongoose database schemas
│   ├── routes/         # API route definitions
│   ├── scheduler/      # Scheduled tasks & reminder management
│   ├── utils/          # Utility functions (date parsing, OAuth setup, etc.)
│   └── server.js       # Entry point for the backend server
│
├── frontend/
│   ├── public/         # Static files (index.html, icons, manifest)
│   ├── src/            # React components and pages
│   └── vite.config.js  # Frontend bundler configuration
│
├── .gitignore           # Git ignore configuration
└── README.md            # Project documentation (this file)

🚀 Getting Started

Step	Command / Instructions
Backend Setup	cd backend && npm install && cp .env.example .env Edit .env with Mongo URI, JWT secret, Gemini API key, Google OAuth credentials npm run dev
Frontend Setup	cd frontend && npm install && cp .env.example .env Edit .env with VITE_API_BASE_URL=http://localhost:4000 npm run dev

---

🔑 Environment Variables

File	Variables
backend/.env	PORT=4000 MONGO_URI=<mongo-uri> JWT_SECRET=<secret> GEMINI_API_KEY=<key> EMAIL=<smtp_email> EMAIL_PASS=<smtp_pass> GOOGLE_CLIENT_ID=<id> GOOGLE_CLIENT_SECRET=<secret> GOOGLE_REDIRECT_URI=http://localhost:4000/api/google/callback
frontend/.env	VITE_API_BASE_URL=http://localhost:4000

---

🛠️ Usage

# Example command
Create a Google Meet with team Friday 5 PM and email boss

📦 Available Scripts

Backend

npm run dev     # Start development server with nodemon
npm start       # Start production server

Frontend

npm run dev     # Start Vite development server
npm run build   # Production build
npm run preview # Preview production build

🤝 Contributing

git checkout -b feature/your-feature
git commit -m "feat: add your feature"
git push origin feature/your-feature

---

© 2025 Happy Semicolon