InsightGraph AI

InsightGraph AI is a production-style multi-agent AI research and analytics platform that transforms unstructured and structured data into reliable, evidence-backed insights. Built with LangGraph, the system orchestrates multiple specialized AI agents that collaborate to analyze documents, perform web research, process datasets, generate visualizations, fact-check information, and produce comprehensive reports.

Live Demo

Live Application: https://insightgraph-ai.streamlit.app/

Features

• Multi-Agent AI workflow powered by LangGraph
• Intelligent PDF document analysis
• Real-time web research using Tavily Search
• CSV data analysis with automated visualizations
• Retrieval-Augmented Generation (RAG)
• ChromaDB vector database for semantic search
• AI-powered fact checking
• Knowledge merging from multiple sources
• Automated report generation
• Export reports for sharing
• Interactive Streamlit interface..

---

Architecture

The application follows a supervisor-agent architecture where a central orchestrator delegates tasks to specialized AI agents.

                    User Query
                         │
                         ▼
                  Supervisor Agent
                         │
 ┌───────────────┬──────────────┬───────────────┐
 ▼               ▼              ▼               ▼
PDF Agent    Web Agent     Data Agent   Visualization
     │             │             │             │
     └─────────────┴─────────────┴─────────────┘
                         │
                  Fact Checker
                         │
                  Knowledge Merger
                         │
                     Reviewer
                         │
                  Report Writer
                         │
                    Final Report

---

AI Agents

Supervisor Agent

Coordinates the workflow and routes tasks to the appropriate specialized agents.

PDF Agent

Extracts and retrieves relevant information from uploaded PDF documents using RAG.

Web Agent

Performs live web research and gathers up-to-date information.

Data Agent

Processes CSV datasets, performs analysis, and extracts key insights.

Visualization Agent

Creates charts and visual representations from structured data.

Fact Checker

Verifies generated information against retrieved evidence.

Knowledge Merger

Combines findings from documents, web research, and data analysis into a unified knowledge base.

Reviewer

Reviews and improves the generated content before report creation.

Report Writer

Produces a structured, comprehensive final report.

Then you can ask follow-up questions about the full project you have built.

---

Technology Stack

AI & LLM

• LangGraph
• LangChain
• OpenAI GPT
• Retrieval-Augmented Generation (RAG)

Vector Database

• ChromaDB

Backend

• Python
• Pandas

Frontend

• Streamlit

Search

• Tavily Search API

Visualization

• Plotly
• Matplotlib

---

Project Structure

InsightGraph-AI/
│
├── agents/
├── graph/
├── services/
├── tools/
├── prompts/
├── utils/
├── frontend/
├── data/
├── exports/
└── streamlit_app.py

---

Installation

git clone https://github.com/efti-dot/InsightGraph-AI

cd InsightGraph-AI

pip install -r requirements.txt

streamlit run frontend/streamlit_app.py

---

Author

MD Iftekhar Hossain

LinkedIn: https://www.linkedin.com/in/md-iftekhar-hossain-196ba32b9/?skipRedirect=true

GitHub: https://github.com/efti-dot