Ex-Apple Engineers Release Push-Activated AI Button

Two former Apple Vision Pro engineers, Chris Nolet and Ryan Burgoyne, unveiled Button: a small, iPod-shuffle–style wearable that hosts a generative AI chatbot and only listens when physically pressed. The device is available for preorder at $179 with shipping planned for December. Button answers out loud or can route audio via Bluetooth to earbuds or smart glasses.

Button is positioned against a string of recent AI wearables that stumbled on privacy, latency, and user value. The founders emphasize press-to-activate privacy as a hard design constraint: Button won’t record continuously, addressing ‘‘the icky’’ problem Nolet described after meeting someone later revealed to have been recording conversations. Wired’s demo reporting stresses the device prioritizes near-instant responses, a direct reaction to critiques that earlier gadgets delivered painfully slow fetch-and-respond experiences.

Both Wired and 9to5Mac highlight the device’s deliberate resemblance to an iPod Shuffle and note the core selling points are immediacy and explicit user consent for listening. Wired observed a demo where Button delivered rapid recommendations; the team claims this latency improvement differentiates it from the Humane AI Pin, which faced criticism for sluggishness. 9to5Mac frames a broader skepticism: why invest in bespoke hardware when existing smartphones can run AI assistants, and prior hardware-first attempts (Humane AI Pin, Rabbit R1) failed to find sustainable product-market fit.

Button crystallizes the current tension between hardware-first UX experiments and app-centric AI deployment. The product raises two practical questions for practitioners building multimodal, low-latency agents: how to architect privacy-by-design activation flows, and whether latency/connection constraints justify specialized edge hardware versus optimized mobile apps or companion devices. Button’s demo claims about responsiveness will be the crucial technical metric to validate its UX argument.

Verify Button’s latency profile under real network conditions, the system architecture (on-device inference vs. optimized cloud calls), and actual privacy guarantees in the final firmware and data-handling policies. Also monitor user adoption signals and reviews after December shipments; that will determine whether this hardware iteration overcomes prior AI-wearable failures.