Grounding AI Chatbots with Full Linux Environment on Docker

 

⚠️ This application is currently under active development

 

Overview

Monadic Chat is a framework designed to create and use intelligent chatbots. By providing a full-fledged Linux environment on Docker to GPT-4 and other LLMs, it allows the chatbots to perform advanced tasks that require external tools for searching, coding, testing, analysis, visualization, and more.

• Documentation

  • English
  • 日本語

• Download Installer

  • MacOS (Apple Silicon/Intel)
  • Windows

 

 

 

 

There are two versions of Monadic Chat. One is a web browser-based app framework developed in this repository. The other is a command line app provided as a RubyGem.

• Monadic Chat (this repository)
• Monadic Chat CLI

Features

Basic Structure

• 🤖 Chat functionality powered by GPT-4 via OpenAI's Chat API
• 👩‍💻 Installable as a GUI application on Mac and Windows using Electron
• 🌐 Usable as a web application in browsers

AI + Linux Environment

• 🐧 Provides a Linux environment (Ubuntu) freely accessible by AI
• 🐳 Tools for LLMs via Docker containers
  • Python (+ pip) for tool/function calls
  • Ruby (+ gem) for tool/function calls
  • PGVector (+ PostgreSQL) for DAG using vector representation
  • Selenium (+ Chrome/Chromium) for web scraping
• 📦 Each container can be managed via SSH
• 📓 Python container can launch Jupyter Notebook

Data Management

• 💾 Export/import conversation data
• 💬 Specify the number of recent messages (active messages) to send to the API
• 🔢 Generate text embeddings from data in PDF files
• 📂 Local data folders are synchronized with Docker containers for seamless interaction

Voice Interaction

• 🎙️ Automatic transcription of microphone input using OpenAI's Whisper API
• 🔈 Text-to-speech functionality for AI assistant responses
• 🗺️ Automatic language detection for appropriate text-to-speech playback
• 🗣️ Choose the language and voice for text-to-speech
• 😊 Enable interactive conversations with the AI agent using speech recognition and text-to-speech

Image and Video Recognition and Generation

• 🖼️ Generate images from text prompts using OpenAI's DALL·E 3 API
• 👀 Analyze and describe the content of uploaded images
• 🎥 Recognize and describe the content and audio of uploaded videos

Configuration and Extension

• 💡 Customize the AI agent's behavior by specifying API parameters and the system prompt
• 💎 Extend functionality using the Ruby programming language
• 🐍 Extend functionality using the Python programming language
• 🌎 Perform web scraping using Selenium

Message Editing

• 📝 Edit previous messages
• 🗑️ Delete specific messages
• 📜 Set roles (user, assistant, system) for new messages

Support for Multiple LLM APIs

• 👥 In addition to OpenAI GPT-4 (default), basic functionality (chat + tool/function call) is also available for:
  • Google Gemini
  • Anthropic Claude
  • Cohere Command R

Managing Conversations as Monads

• ♻️ Manage (invisible) conversation state by obtaining additional responses from LLM and updating values in a predefined JSON object

Author

Yoichiro HASEBE
yohasebe@gmail.com

License

The gem is available as open source under the terms of the MIT License.