This project provides a proof-of-concept for an AI agent's "brain," enabling it to recall information from memory or learn from the web when it encounters unknown queries. It's designed for developers and researchers experimenting with AI agent architectures and memory management.

How It Works

The agent utilizes a vector database (ChromaDB by default) to store and retrieve memories. When queried, it first checks its memory. If the answer isn't found, it browses the web, learns the information, and then saves it to its memory store for future recall.

Quick Start & Requirements

• Install dependencies: npm i
• Run agent: npm run agent
• Prerequisites: Node.js, OpenAI API Key, optional SerpAPI Key.
• Local ChromaDB instance requires Docker.
• Demo: https://ai-brainstore.vercel.app/

Highlighted Details

• Vector database integration for memory storage.
• Web browsing capability for knowledge acquisition.
• Configurable memory review and model selection.

Maintenance & Community

• Project maintained by mckaywrigley.
• Contact available via Twitter.

Licensing & Compatibility

• License not specified in README.
• Compatibility for commercial use is undetermined.

Limitations & Caveats

This is explicitly a proof-of-concept and may not be production-ready. The reliance on external API keys (OpenAI, SerpAPI) is a significant dependency.