At the Edge - What it Really Takes to Build AI Wearables

Session details:

Though AI wearables have gained immense traction in the past few years, the promise of truly ambient, hands-free intelligence is not guaranteed. To do so, we must overcome the real constraints that hardware, software, and interaction design have historically placed on these devices. Let's explore how.

Join Austin, creator and leader of Watch AI for Google Pixel, as we move beyond general concepts in this technical deep-dive of the unique challenges and solutions required to run powerful, personalized AI directly on the wrist. We will:

• Examine the he failure points of previous wearable generations to understand the unique factors relevant to edge computing
• Analyze the real-world engineering trade-offs between on-device and cloud processing
•  Explore software architectures and hardware co-optimization techniques that enable highly efficient on-device inference, and define the relevant factors to consider on-device AI.

Attendees will walk away with practical deployment strategies, using Pixel Watch and it's deployment of the first on-Watch LLM as a case study