🤖 WorkFlowGen — Generative AI Document Intelligence Platform

An End-to-End Agentic RAG System for Intelligent Document Processing and Analysis

Key Features • Tech Stack • Architecture • Installation • Usage

WorkFlowGen moves beyond traditional chatbots by implementing a Semantic Router powered by Google Gemini 2.0 Flash.
It intelligently classifies user intent (QA vs. Summarization) and routes requests to specialized agents using a LangGraph State Machine.

---

🚀 Key Features

Feature	Description
🧠 Agentic Workflow	Replaces linear logic with a cyclic LangGraph state machine for decision-making.
🎯 Semantic Routing	Uses structured JSON output to distinguish between "help/critique" (QA) and "overview" (Summary) requests without brittle keywords.
⚡ Structured Output	Enforces strict JSON schemas on the LLM to prevent routing errors and hallucinations.
🔍 Smart Retrieval	Uses pgvector for similarity search to fetch precise context from documents.
📄 Async Ingestion	Handles PDF/Text parsing and vectorization asynchronously using FastAPI.
📊 Modern UI	Real-time, responsive dashboard built with Next.js 16, Tailwind CSS, and Lucide Icons.

---

🛠️ Tech Stack

Frontend	Backend	Data & AI	Deployment
Next.js 16, Tailwind CSS, Lucide Icons	FastAPI, LangGraph	pgvector, Google Gemini 2.0 Flash	Docker, PostgreSQL

---

🏗️ Architecture

The system follows a cyclic graph architecture rather than a linear chain:

flowchart TD
    A[User Input] --> B[FastAPI Endpoint]
    B --> C{LangGraph Workflow}

    subgraph Agent_Brain [Agent Brain]
        C --> D[Retriever Node]
        D --> E[Semantic Router]

        E -->|Intent: QA| F[QA Generator]
        E -->|Intent: SUMMARY| G[Summary Generator]
    end

    F --> H[Final Response]
    G --> H

Loading

---

💻 Local Installation Guide

Follow these steps to set up and run the WorkFlowGen Agentic Stack on your local machine.

---

🧩 1. Prerequisites

Before installation, make sure you have the following:

Tool	Version	Purpose
🐳 Docker Desktop	Latest	To run PostgreSQL (with pgvector)
🐍 Python	3.10+	Backend (FastAPI + LangGraph)
💚 Node.js	18+	Frontend (Next.js 16 + Tailwind CSS)

---

📥 2. Clone the Repository

git clone https://github.com/Prachet-Dev-Singh/WorkFlowGen.git
cd WorkFlowGen

---

🗃️ 3. Database Setup (PostgreSQL + pgvector via Docker)

Spin up the database container:

docker compose -f infra/docker-compose.yml up -d

✅ Verify that your database is running:

docker ps

This launches a PostgreSQL instance with pgvector extension pre-installed for efficient semantic search.

---

⚙️ 4. Backend Setup (FastAPI + LangGraph)

Navigate to the backend directory and create a virtual environment:

cd backend
python -m venv venv

Activate the environment:

# Windows (Git Bash)
source venv/Scripts/activate

# Mac/Linux
source venv/bin/activate

Install dependencies:

pip install -r requirements.txt

🔑 Configure Environment Variables

Create a .env file inside the backend folder:

touch .env

Paste the following and replace YOUR_GEMINI_API_KEY_HERE:

# Local Docker Database
DATABASE_URL=postgresql+asyncpg://admin:password123@localhost:5432/workflowgen_db

# Google Gemini API Key
GEMINI_API_KEY=YOUR_GEMINI_API_KEY_HERE

🗄️ Initialize the Database Schema

python init_db.py
# ✅ Tables created successfully!

🚀 Start the Backend Server

uvicorn app.main:app --reload

Backend runs on → http://localhost:8000

---

🎨 5. Frontend Setup (Next.js + Tailwind CSS)

Open a new terminal and navigate to the frontend directory:

cd frontend
npm install
npm run dev

Frontend runs on → http://localhost:3000

---

🧪 6. Testing the Application

1. Open http://localhost:3000 in your browser.
2. Upload a document (PDF or text) using the left upload panel.
3. Try different queries:
   • 📝 “Summarize this document.” → Routed to Summary Agent
   • 💬 “How can I improve this email?” → Routed to QA/Critique Agent
4. Observe the backend terminal logs for real-time routing and reasoning steps.

---

🧹 7. Stopping the Containers

When done, stop Docker containers safely:

docker compose -f infra/docker-compose.yml down

---

🧰 Quick Recap

Component	Tech	Port	Description
🧠 Backend	FastAPI + LangGraph	8000	Handles semantic routing and agent logic
🖥️ Frontend	Next.js + Tailwind	3000	Interactive UI for document upload and analysis
🗃️ Database	PostgreSQL + pgvector	5432	Vector similarity storage

---

🎉 You’re all set!
WorkFlowGen is now fully functional on your local machine — explore, extend, and build agentic intelligence on your own data!

---

🧪 How to Use

Once both servers are running:

1. Open http://localhost:3000
2. Use the Upload Panel to add PDFs or .txt files.
3. Enter natural language queries:
   • 🧠 “Summarize this document.” → Routed to Summary Agent
   • 💬 “What is the main argument of section 2?” → Routed to QA Agent
4. Watch the backend logs to see:
   • Intent classification (QA / SUMMARY)
   • Node transitions in the LangGraph
   • Final structured JSON response

---

💡 Example Queries

Type	Example	Agent Invoked
Summary	"Give me a short overview of this file"	📝 Summary Generator
Critique	"How can I improve this paragraph?"	💡 QA Generator
Clarification	"Explain section 3 in simple terms"	💡 QA Generator

---

📊 Debug View

In the backend terminal, you’ll see live logs like:

[Router] Intent detected: SUMMARY
[SummaryAgent] Fetching context from pgvector...
[SummaryAgent] Generating overview response...
[Graph] Returning final output to user.

This shows how the LangGraph state machine routes and executes agents dynamically.

---

📄 License

This project is licensed under the MIT License.

---