Hey, this guy again? Yes! Because a LOT has happened in AI in the past 3 years. Suddenly we've have huge multi-modal LLMs that have a pretty good visual understanding of the world; we have generative AI models that can just spit out freshly created synthetic audio and image data; and we're starting to see the first foundation models for sensor data. So... how does that apply to you and your devices?! In this session we'll explore the bleeding edge of LLMs and Generative AI; how you can use large models to train small models (that will run on your IoT device), how to set up effective cascades from device to the cloud for further analysis (without burning through your battery in a day); and how you can use GenAI to label or generate data a lot faster.