DocAgent - Intelligent Multi-Document Processor

A production-grade agentic document processing system built with LangGraph, Groq Llama 3.3 70B, and Langfuse observability. Upload invoices, expense reports, or booking confirmations, or even your resume - the agent classifies, extracts, validates, and lets you chat with your documents.

---

Live Demo

URL: Chat with my agent here

• Note : No login required, may take a minute to wake up (Render free tier limits on inactivity)

---

Langfuse Observability

Every document processing run and chat message is traced in your Langfuse dashboard:

• Full node-by-node traces (load - classify - extract - validate - respond)
• LLM call inputs/outputs with latency
• Per-user session tracking
• Anomaly rates and risk distributions

---

Example Questions

After uploading documents:

• "What is the total amount on the invoice?"
• "Are there any anomalies across all documents?"
• "Who is the traveler on the booking confirmation?"
• "Summarize all documents in one paragraph"

---

Tech Stack

Component	Technology
Agent framework	LangGraph 0.2+
LLM	Groq Llama 3.3 70B (free tier)
Document loading	LangChain PyPDFLoader
Observability	Langfuse cloud (free tier)
UI	Gradio 4
Backend	FastAPI + Uvicorn
Deployment	Render (free tier)

---

Quick Start (Local)

1. Clone & install

git clone https://github.com/EESHAK02/doc-processor
cd doc-processor
pip install -r requirements.txt

2. Set environment variables

Edit .env with your keys

You need:

• Groq API key - free at https://console.groq.com
• Langfuse keys - free at https://cloud.langfuse.com (create a project, copy Public + Secret keys)

3. Run

python main.py

Open http://localhost:7860 in your browser.

---

Deploy to Render (Free Public Link)

1. Push this repo to GitHub
2. Go to https://render.com -> New -> Web Service
3. Connect your GitHub repo
4. Set these environment variables in Render dashboard:
   • GROQ_API_KEY
   • LANGFUSE_PUBLIC_KEY
   • LANGFUSE_SECRET_KEY
   • LANGFUSE_HOST → https://cloud.langfuse.com
5. Build command: pip install -r requirements.txt
6. Start command: python main.py
7. Deploy - Render gives you a public https://your-app.onrender.com URL