Skip to content

Latest commit

 

History

History
106 lines (76 loc) · 4.16 KB

README.md

File metadata and controls

106 lines (76 loc) · 4.16 KB
Raw

Task mAIstro

Managing tasks effectively is a universal challenge. Task mAIstro is an AI-powered task management agent that combines natural language processing with long-term memory to create a more intuitive and adaptive experience. This repo can be used to deploy Task mAIstro and interact with it through text or voice!

Key features:

  • Natural conversation through text and / or voice to update or add tasks
  • Adaptive learning of your management style and preferences
  • Persistent memory of tasks, context, and preferences
  • Flexible deployment options - local or hosted

audio_task_maistro

Quickstart

  1. Populate the .env file:
$ cp .env.example .env

  1. Download the LangGraph Studio desktop app for Mac here.

  2. Load this repository as a project in LangGraph Studio.

  3. Start chatting with the task mAIstro through the text interface in LangGraph Studio.

  4. Experiment with voice UX in ntbk/audio_ux.ipynb.

Task mAIstro Application

Architecture

Task mAIstro leverages LangGraph to maintain three memory types:

  1. ToDo List Memory

    • Task descriptions and deadlines
    • Time estimates and status tracking
    • Actionable next steps

  2. User Profile Memory

    • Personal preferences and context
    • Work/life patterns
    • Historical interactions

  3. Interaction Memory

    • Task management style
    • Communication preferences
    • Organizational patterns

The schema for each memory type as well as the graph flow is defined in task_maistro.py. The graph flow is orchestrated by a central task_maistro node that:

  • Chooses to update one of the three memories based on the user's input
  • Uses tool calling with the Trustcall library to update the chosen memory type

📚 Learning Resources

Want to learn more? Check out Module 5 of our LangGraph Course:

  • All notebooks are available here
  • All notebooks have accompanying videos

Deployment Options

Task mAIstro offers three flexible deployment paths:

1. LangGraph Studio (Recommended for quickstart)

The fastest way to get started is with LangGraph Studio:

Screenshot 2024-11-20 at 7 42 35 PM

2. LangGraph CLI

For developers who prefer the command line:

  • Use the LangGraph CLI
  • Deploy locally with a single command
  • Full control over deployment configuration

3. LangGraph Cloud

For production deployments:

  • Deploy to LangGraph Cloud
  • Manage through your LangSmith account
  • Access via LangGraph Studio web UI

🚀 Learning Resources

Want to learn more about deployments? Check out Module 6 of our LangGraph Course

  • All notebooks are available here
  • All notebooks have accompanying videos

Voice Interface

Task mAIstro supports voice interactions using:

Setup

  1. Install FFmpeg (required for ElevenLabs):
# macOS
brew install ffmpeg
  1. In audio_ux.ipynb, connect to your deployment using the URL endpoint:
  • Studio: Found in Studio UI
  • CLI: Printed to console (e.g., typically http://localhost:8123)
  • Cloud: Available in the LangSmith Deployment page