Open AI Home Assistant

In the last few years AI advancements have exploded, with models like GPT-4 (now 5) transforming how we interact with technology. This project was to learn how to leverage and integrate OpenAI's powerful models into a Home Assistant setup, enabling voice-activated control and dynamic content creation for smart home environments. My short term goal with this project was to learn more about API integration, integration of publically available services, and to have fun! The long term goal is to with user driven routine creation through voice commands. I had originally wanted to use this as a voice assistant in my campervan - the use case I sold to my partner was being able to turn the van heating on remotely from the top of a mountain, and returning to a nice toasty van!

Learning Outcomes

1. Integration with OpenAI GPTs for completions and generating responses
2. Basic TTS, routines and wake work recogntion
3. Learned about Flask authentification and developed basic auth priciples
4. Explored features available to OpenAI Assistants
5. Basic function calling from AI Assistants

Project Architecture

• Core Features:
  • friday.py is the main entry point for this project. It checks the routines folder and updates the functions available to call; then listens for the wake command "Hey Friday"
• Extensibility:
  • This project was very eye-opening to undertake - since I put it down, I know that there have been many more developer-centric enhancements that should be revisited. Next steps for this project are to integrate more deeply with OpenAI agents, and to explore function calling with GPT models to enable more dynamic and context-aware interactions.

Explore the Code

• View on GitHub