AInalyst is an advanced AI-powered research and analysis assistant that transforms complex queries into comprehensive reports and professional presentations. Built with Node.js and leveraging cutting-edge AI models (gpt-oss-20b and gpt-oss-120b), AInalyst automates the research process, saving hours of manual work while delivering high-quality, well-structured insights.

• Intelligent Research: Automatically breaks down complex queries into sub-questions for thorough investigation
• Multi-Source Information Gathering: Utilizes multiple web search providers for comprehensive data collection
• AI-Powered Analysis: Employs advanced language models to synthesize information into coherent insights
• Professional Report Generation: Creates well-structured markdown reports with proper citations
• Presentation Creation: Automatically generates PowerPoint presentations from research findings
• Dual Interface: Command-line interface for quick tasks and web interface for interactive research
• Research History: Maintains a history of all research investigations for easy access
• Export Options: Download results in both markdown and PowerPoint formats

• Literature reviews and topic exploration
• Thesis and dissertation support
• Academic paper drafting assistance

• Market analysis and competitive research
• Industry trend identification
• SWOT analysis generation

• Blog post and article research
• Social media content ideation
• Presentation material development

• Topic deep-dives and knowledge acquisition
• Study material organization
• Research skill development

• Node.js (v14 or higher)
• An API key (for gpt-oss models).
• An OpenAI compatible Base URL (for gpt-oss models, by default is https://router.huggingface.co/v1).
• An OpenAI Model name (the default name is openai/gpt-oss-20b).
• Internet connection for web searches

1. Clone the repository:

   git clone <repository-url>
   cd ainaylst

2. Install dependencies:

   npm install

3. Set up your API keys:

   • Copy the .env.example file to .env
   • Add your API keys to the .env file:

     OPENAI_API_KEY=sk-your-openai-api-key-here
     OPENAI_BASE_URL=your-openai-base-url-here
     OPENAI_MODEL=your-openai-model-here
     MAX_TOKENS=your-max-tokens-here (default is 5000)
     
     # For Tavily (optional)
     TAVILY_API_KEY=tvly-your-tavily-api-key-here
     # For Exa (optional)
     EXA_API_KEY=your-exa-api-key-here
     
     # Results size
     MAX_RESULTS=your-max-results-here (default is 5)

Run the application in interactive CLI mode:

npm run cli

Then enter your research query when prompted.

Run the application with a direct query:

npm start "Your research query here"

Or:

node src/index.js "Your research query here"

Run the web server:

npm run web

Then open your browser to http://localhost:3000

npm start "What are the environmental impacts of electric vehicles compared to traditional cars?"

npm start "Analyze the key factors driving the growth of renewable energy adoption worldwide"

You can select which retriever to use in several ways:

1. Set the RETRIEVER environment variable in your .env file:

   RETRIEVER=exa

2. Use the --retriever flag when running the application:

   npm start -- --retriever exa "Your research query here"

3. When using the CLI mode, you'll be prompted to select a retriever.

Available retrievers:

• tavily (default)
• exa

1. Research Planning: AInalyst analyzes your query and breaks it down into targeted sub-questions
2. Information Gathering: For each sub-question, it searches the web using multiple providers for relevant information
3. Content Synthesis: Advanced AI models synthesize the gathered information into coherent, well-structured answers
4. Report Generation: All findings are compiled into a comprehensive markdown report with proper citations
5. Presentation Creation: The markdown report is automatically converted into a professional PowerPoint presentation

src/
├── index.js          # Main entry point
├── cli.js            # CLI interface
├── server.js         # Web server
├── public/           # Web interface files
│   ├── index.html    # Main HTML file
│   ├── styles.css    # Styles
│   └── script.js     # Client-side JavaScript
├── retrievers/       # Web search functionality
│   ├── BaseRetriever.js     # Base retriever class
│   ├── TavilyRetriever.js   # Tavily search provider
│   └── ExaRetriever.js      # Exa search provider
├── skills/           # Research and report generation skills
└── utils/            # Utility functions and configuration

The web interface provides a modern, ChatGPT-like experience with the following features:

1. Chat-based Interaction: Enter your research questions in a conversational interface
2. Real-time Progress Updates: See the status of your research as it progresses through different stages
3. Research History: Access previous research investigations from the sidebar
4. Results Display: View formatted research reports in the browser
5. Presentation Mode: View your research as an interactive slide presentation
6. Download Options: Download both markdown reports and PowerPoint presentations
7. Responsive Design: Works seamlessly on desktop and mobile devices

• OPENAI_API_KEY: Your OpenAI API compatible key (required)
• OPENAI_BASE_URL: Your OpenAI API compatible base URL (required, by default is https://router.huggingface.co/v1)
• OPENAI_MODEL: Your OpenAI API compatible model name (required or default: openai/gpt-oss-20b)
• MAX_TOKENS: Max tokens (optional or default: 5000)
• MAX_RESULTS: Max results for web search (optional or default: 5)
• TAVILY_API_KEY: Your Tavily API key (optional, but recommended)
• EXA_API_KEY: Your Exa API key (optional)
• RETRIEVER: Default retriever to use (tavily or exa)
• PORT: Port for the web server (default: 3000)

You can customize the output by modifying the templates in the skills directory:

• ReportGenerator.js: Controls the structure and formatting of markdown reports
• PresentationGenerator.js: Controls the creation of PowerPoint presentations

Generated reports are saved as well-formatted markdown files with:

• Clear section headings
• Proper citation formatting
• Logical content organization
• Slide separators for presentation creation

Automatically generated presentations include:

• Professional slide layouts
• Proper content sizing and positioning
• Visual separation of topics
• Consistent styling throughout

The application is designed to be easily extensible with new retrievers:

1. Create a new class that extends BaseRetriever
2. Implement the required search and scrapeContent methods
3. Add your retriever to the createRetriever method in ResearchSkill.js
4. Update the CLI and environment variable handling if needed

While AInalyst is powerful, it has some limitations:

• Research quality depends on the availability of relevant web content
• Generated content should be verified for accuracy, especially for critical applications
• API usage costs may accumulate with extensive use
• No offline functionality (requires internet connection)

API Key Errors: Ensure your API keys are correctly set in the .env file and have sufficient credits.

Network Issues: Verify your internet connection and firewall settings if research fails to complete.

Contributions are welcome! Feel free to submit issues or pull requests for:

• Bug fixes
• New features
• Performance improvements
• Documentation enhancements
• Additional retriever implementations

This project is licensed under the MIT License.

AInalyst builds upon the concepts and methodologies of AI research assistants, combining the power of large language models with efficient information retrieval systems to create a practical tool for researchers, students, and professionals.