ESG Reporter - Multi-Agent AI Research System

A sophisticated multi-agent AI system built with LangGraph that orchestrates specialized agents to conduct comprehensive ESG (Environmental, Social, and Governance) research. The system combines web search, academic research, and document analysis to generate detailed ESG reports.

🎯 Ready-to-Use: Includes 30+ curated ESG documents from major corporations, frameworks, and sustainability reports.

🌟 Features

• Multi-Agent Architecture: 6 specialized AI agents working collaboratively
• Comprehensive Research: Web search, academic papers, and document analysis
• Azure Integration: Powered by Azure OpenAI and Azure AI Search
• Interactive Interface: Beautiful Streamlit web application
• Document Processing: Automated PDF indexing with vector search
• Real-time Visualization: Watch agents work in real-time

🤖 AI Agents

1. Lead Agent - Orchestrates research and coordinates other agents
2. Academic Research Agent - Searches arXiv and academic databases
3. Web Search Agent - Conducts comprehensive web research using Brave Search
4. Data Search Agent - Performs vector search on indexed ESG documents
5. Verification Agent - Fact-checks and validates research findings
6. Synthesis Agent - Combines findings into comprehensive reports

🚀 Quick Start

Prerequisites

• Python 3.8+
• Azure OpenAI API access
• Azure AI Search service
• Brave Search API key (free)

Installation

1. Clone the repository

   git clone https://github.com/DarshKodwani/ESGreporter.git
   cd ESGreporter

2. Set up virtual environment

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies

   pip install -r requirements.txt

4. Configure environment variables

   cp .env.example .env
   # Edit .env with your API keys and endpoints

5. Set up Azure AI Search index

   # Create the search index using the provided schema
   # See setup instructions below

6. Index your ESG documents

   # The repository includes 30+ ESG PDF documents in the ESG_docs folder
   # Run the indexing script to upload them to Azure AI Search
   python add_documents_to_index.py

7. Launch the application

   streamlit run streamlit_app.py

⚙️ Configuration

Environment Variables

Create a .env file with the following variables:

# Azure OpenAI Configuration
AZURE_OPENAI_ENDPOINT=your_azure_openai_endpoint
AZURE_OPENAI_API_KEY=your_azure_openai_api_key
AZURE_OPENAI_API_VERSION=2024-02-15-preview
AZURE_OPENAI_DEPLOYMENT_NAME=gpt-4o

# Azure AI Search Configuration
AZURE_SEARCH_ENDPOINT=your_search_endpoint
AZURE_SEARCH_KEY=your_search_key

# Brave Search API
BRAVE_SEARCH_API_KEY=your_brave_api_key

# Azure OpenAI Configuration - Embeddings
AZURE_OPENAI_EMBEDDINGS_ENDPOINT=your_embeddings_endpoint
AZURE_OPENAI_EMBEDDINGS_API_KEY=your_embeddings_api_key
AZURE_OPENAI_EMBEDDINGS_DEPLOYMENT_NAME=text-embedding-3-large

Azure AI Search Setup

1. Create the search index using Azure Portal, CLI, or REST API:

   # Using Azure CLI
   az search index create --service-name your-search-service --index-definition azure-index-schema.json

2. Or use the provided schema in azure-index-schema.json to create the index manually

3. Configure your Azure services:

   • Azure OpenAI: Deploy GPT-4 and text-embedding-3-large models
   • Azure AI Search: Create search service and index with vector capabilities
   • Brave Search: Get free API key from Brave Search API

4. Update environment variables in .env with your actual service endpoints and keys

📁 Project Structure

ESGreporter/
├── agents.py                 # AI agent definitions and logic
├── tools.py                  # Search tools and utilities
├── workflow.py              # LangGraph workflow orchestration
├── streamlit_app.py         # Web interface
├── add_documents_to_index.py # Document indexing script
├── azure-index-schema.json  # Azure AI Search index schema
├── requirements.txt         # Python dependencies
├── setup.py                # Package setup configuration
├── firstTimeSetup.sh       # Initial setup script
├── .env.example            # Environment variables template
├── .gitignore              # Git ignore rules
├── ESG_docs/               # ESG PDF documents (30+ files)
│   ├── 2025-Microsoft-Environmental-Sustainability-Report.pdf
│   ├── Apple_Environmental_Progress_Report_2025.pdf
│   ├── 241213-hsbc-green-bonds-report-2024.pdf
│   └── ... (27 more ESG documents)
└── README.md               # This file

🎯 Usage

1. Launch the web interface: streamlit run streamlit_app.py
2. Enter your research query in the sidebar
3. Watch the agents work in real-time through the cinematic interface
4. Download results as PDF reports
5. Explore cached results from previous research sessions

🔧 Advanced Features

Document Indexing

The system includes a comprehensive collection of ESG documents for enhanced research:

# The ESG_docs folder contains 30+ PDF documents including:
# - Microsoft Environmental Sustainability Reports (2020, 2022, 2024, 2025)
# - Apple Environmental Progress Reports (2023, 2025)
# - HSBC ESG Reviews and Green Bonds Reports (2021-2024)
# - Various corporate sustainability reports and ESG frameworks

# Run the indexing script to upload documents to Azure AI Search
python add_documents_to_index.py

Document Collection Includes:

• Corporate ESG reports from Microsoft, Apple, HSBC, and other major companies
• Environmental sustainability reports and climate disclosures
• Green bonds and sustainable finance documents
• ESG framework documents and standards
• Social impact and governance reports

Custom Agent Configuration

Modify agent behaviors in agents.py:

• Adjust system prompts
• Add new search capabilities
• Customize response formats

Workflow Customization

Enhance the agent orchestration in workflow.py:

• Add new agent types
• Modify execution flow
• Implement parallel processing

📊 Technical Architecture

• Framework: LangGraph for agent orchestration
• LLM: Azure OpenAI GPT-4
• Vector Search: Azure AI Search with text-embedding-3-large
• Web Search: Brave Search API
• Academic Search: arXiv API
• Frontend: Streamlit with custom CSS
• Document Processing: PyPDF2 with intelligent chunking

🤝 Contributing

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• Built with LangGraph for agent orchestration
• Powered by Azure OpenAI
• Search capabilities via Brave Search API
• Academic research through arXiv API

📞 Support

For questions and support:

• Open an issue on GitHub
• Review the comprehensive code documentation
• Check the Azure AI Search and OpenAI documentation for service setup

---

Made with ❤️ for ESG research and sustainable investing