CleverBee is a groundbreaking research assistant that harnesses the power of artificial intelligence, specifically utilizing advanced Large Language Models (LLMs) like Claude and Gemini. This innovative tool is designed to streamline the research process by effectively collaborating with the user, performing web browsing tasks, extracting relevant content, and synthesizing findings based on specified research topics. The application is built using Python and leverages Playwright for efficient web browsing, alongside Chainlit to provide an interactive user interface.

Key Features

• Interactive Web UI: CleverBee employs Chainlit to create a user-friendly and interactive web interface, making research more accessible and engaging.
• MCP Tool Support: Through the utilization of the Model Context Protocol (MCP), CleverBee integrates with various external tools, enhancing its functionality.
• Multi-LLM Research: CleverBee smartly employs distinct and configurable LLMs tailored for specific tasks. For instance, it uses Gemini 2.5 Pro for planning and generating final reports, Gemini 2.5 Flash for analyzing research progress and determining subsequent actions, and Gemini 2.0 Flash for summarizing web content to aid in planning further steps.
• Automated Web Browsing: The application utilizes Playwright to efficiently search the web and extract HTML content, ensuring users have access to the most current information available.
• Content Processing: CleverBee adeptly cleans HTML content, converting it into Markdown format before summarization, which simplifies the data for users.
• Integrated Token Tracking: To assist users in managing costs associated with LLM calls, CleverBee includes a feature that monitors token usage and provides cost estimates.
• Highly Configurable: Users have the ability to manage settings through a configuration file (config.yaml), allowing for a personalized user experience.
• Modular LLM Clients: The platform supports various providers, including Gemini, Claude, and local models via the llama-cpp-python library, providing flexibility for users.
• LLM Caching: CleverBee employs a caching mechanism (NormalizingCache, based on SQLite) to optimize performance and reduce operational costs.

What CleverBee Is Not

It is crucial to note that CleverBee is not intended to replace deep domain expertise. Despite its advanced AI capabilities, users are advised to thoroughly verify the sources of information, as AI can occasionally produce inaccuracies or 'hallucinations' even when guided by explicit instructions.

System Compatibility

• macOS: CleverBee is fully tested and supported on macOS, accommodating both Intel and Apple Silicon through Rosetta 2.
• Linux: The application is also fully supported on Linux systems, with built-in NVIDIA GPU detection and optimization for local models.
• Windows: While CleverBee offers limited support on Windows, it can still function via the Windows Subsystem for Linux (WSL).

Quick Start Guide

For users eager to get started with CleverBee, the setup process is straightforward:

1. Clone the repository using the command: git clone http