README.md

metadata

title: Second Brain AI Assistant
emoji: 🧠
colorFrom: blue
colorTo: purple
sdk: gradio
sdk_version: 5.12.0
app_file: app.py
pinned: false
license: mit

Second Brain AI Assistant

A production-ready AI assistant that can answer questions about your documents using RAG (Retrieval-Augmented Generation).

Features

• Document Q&A: Ask questions about your documents
• Source Attribution: See which documents were used for each answer
• Clean UI: Professional interface with proper formatting
• Real-time Processing: Get answers instantly
• Tool Usage Display: See which tools were used to generate responses

Usage

1. Enter your question in the text box
2. Click "Ask" to get an AI-powered answer
3. View sources and tools used in the response
4. Use the debug section to see raw responses

Example Queries

• "What pricing objections have been raised?"
• "What messaging is resonating with prospects?"
• "What concerns have prospects raised with regards to product?"
• "What has resonated with prospects based on the meeting transcripts?"

Configuration

This space uses the following environment variables:

• OPENAI_API_KEY: Your OpenAI API key
• MONGODB_URI: MongoDB connection string
• MONGODB_DATABASE_NAME: Database name (default: second_brain_course)
• MONGODB_COLLECTION_NAME: Collection name (default: rag)
• COMET_API_KEY: Comet ML API key for tracking
• COMET_PROJECT: Project name (default: second_brain_course)
• RETRIEVER_CONFIG_PATH: Path to retriever config (default: configs/compute_rag_vector_index_openai_contextual_simple.yaml)

Architecture

• RAG Pipeline: Uses MongoDB for document storage and retrieval
• Embeddings: OpenAI text-embedding-3-small for document embeddings
• LLM: GPT-4o-mini for response generation
• UI: Custom Gradio interface with enhanced formatting
• Tools: MongoDB retriever and final answer tools

Local Development

# Install dependencies
uv sync

# Run the agent
make run_agent_app

# Or run directly
python app.py

License

MIT License