DeepResearch

A Java Spring Boot service that provides AI-powered research capabilities using LangChain4j and various LLM providers (Ollama, Groq, OpenAI, etc.). It autonomously generates research queries, performs web searches, and creates comprehensive summaries.

✨ Features

• AI-Powered Research: Uses LangChain4j with multiple LLM providers
• Multiple Search Backends: DuckDuckGo, Tavily, Perplexity support
• Beautiful Web UI: Professional markdown rendering with syntax highlighting
• Real-time Feedback: Loading indicators and status updates
• Flexible Configuration: Environment variable support
• Comprehensive Debug Logging: Detailed logging for troubleshooting
• Responsive Design: Works on desktop and mobile

📝 Sample Report

• An sample report generated by this tool on the topic of "Cybersecurity industry deep dive" is available at research_report_1.md
• Another one on the topic of "Explain sorting algorithms with examples in java" is available at research_report_2.md

Research Topic Ideas

• "Latest advances in quantum computing"
• "2024 FDA approved cancer drugs"
• "Climate change mitigation strategies"
• "Machine learning applications in healthcare"

Output Information

The application generates comprehensive markdown reports with:

• Executive summary
• Key findings
• Sources and references
• Formatted tables and lists
• Code examples (if applicable)

🚀 Quick Start

Prerequisites

• Java 21 or higher
• Maven 3.6+
• An LLM provider (OpenAI, Inception, Anthropic, Ollama, Groq, etc.)

1. Clone and Build

git clone <repository-url>
cd DeepResearch
mvn clean package

2. Configure Environment Variables

Default: Use Inception (for speed)

# API key is the only thing needed, since others are defaults 
export RESEARCH_API_KEY=your_inception_api_key

# These are not needed but merely mentioned here for completeness
export RESEARCH_LLM_PROVIDER=inception
export RESEARCH_MODEL_NAME=mercury-coder
export RESEARCH_BASE_URL=https://api.inceptionlabs.ai/v1

Option A: Use Ollama (Local)

# Install Ollama
curl -fsSL https://ollama.ai/install.sh | sh

# Pull a model
ollama pull llama2

# Set environment variables
export RESEARCH_LLM_PROVIDER=ollama
export RESEARCH_MODEL_NAME=llama2
export RESEARCH_BASE_URL=http://localhost:11434

Option B: Use Groq

export RESEARCH_LLM_PROVIDER=groq
export RESEARCH_MODEL_NAME=mixtral-8x7b-32768
export RESEARCH_BASE_URL=https://api.groq.com/openai/v1
export RESEARCH_API_KEY=your_groq_api_key

Option C: Use OpenAI

export RESEARCH_LLM_PROVIDER=openai
export RESEARCH_MODEL_NAME=gpt-4
export RESEARCH_BASE_URL=https://api.openai.com/v1
export RESEARCH_API_KEY=your_openai_api_key

3. Run the Application

mvn spring-boot:run

4. Access the Web Interface

Open your browser to: http://localhost:8080

🎨 Web Interface

The application includes a beautiful, responsive web interface with:

• Markdown rendering with syntax highlighting
• Real-time loading indicators
• Error handling and feedback
• Copy buttons for code blocks
• Mobile-responsive design

🔧 Configuration

Environment Variables

Variable	Description	Default
RESEARCH_LLM_PROVIDER	LLM provider (ollama, groq, openai)	ollama
RESEARCH_MODEL_NAME	Model name to use	llama2
RESEARCH_BASE_URL	Base URL for LLM API	http://localhost:11434
RESEARCH_API_KEY	API key for cloud providers
MAX_RESEARCH_LOOPS	Number of research iterations	3
SEARCH_API	Search backend (duckduckgo, tavily, perplexity)	duckduckgo
MAX_TOKENS	Max tokens per source	1000

Application Properties

The application uses Spring Boot configuration. You can override settings in application.yml.

📊 API Endpoints

Research API

• POST /api/research/conduct - Start research
• GET /api/research/health - Health check
• GET /api/research/debug - Debug information

Example Usage

# Start research
curl -X POST http://localhost:8080/api/research/conduct \
  -H "Content-Type: application/json" \
  -d '{"topic":"Latest advances in quantum computing"}'

# Health check
curl http://localhost:8080/api/research/health

# Debug info
curl http://localhost:8080/api/research/debug

🐛 Debug Mode

Enable debug logging by setting:

export LOGGING_LEVEL_COM_EXAMPLE=DEBUG

🧪 Testing

Manual Testing

1. Start the application
2. Open http://localhost:8080
3. Enter a research topic
4. Click "Run Research"
5. Watch the loading indicator and results

API Testing

# Test with a simple topic
curl -X POST http://localhost:8080/api/research/conduct \
  -H "Content-Type: application/json" \
  -d '{"topic":"What is artificial intelligence?"}'

🔍 Troubleshooting

Common Issues

Research takes too long

• Check your LLM provider connection
• Reduce MAX_RESEARCH_LOOPS if needed
• Ensure your API key is valid

No results appear

• Check browser console for errors
• Verify environment variables
• Test with a simple topic first

Debug Commands

# Check configuration
curl http://localhost:8080/api/research/debug

# Test connection
curl http://localhost:8080/api/research/health

🤝 Contributing

1. Fork the repository
2. Create a feature branch
3. Add comprehensive logging
4. Test with multiple LLM providers
5. Submit a pull request

📄 License

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

🙏 Acknowledgments

• LangChain4j for AI integration
• Marked.js for markdown rendering
• Prism.js for syntax highlighting
• Spring Boot for the framework