I’ve been annoyed for years by those tiny interruptions: need a quick conversion, calc, fact-check, alarm, or translation -> fish out phone -> unlock -> app -> 10-15 seconds of flow broken. Happens 20-50 times a day for me.
So I built Hollow, a small wrist device that’s basically a fast voice AI, like a mini J.A.R.V.I.S. Tap the screen, speak query, get spoken answer in <2 seconds average. Screen only shows time and responses, no notifications.
Current prototype (Hollow 0W):
• Around 1.5-2.5s end-to-end latency in good conditions
• Cloud-based processing
• Battery lasts 2-3 days with moderate use
• Privacy-focused: Use Groq to run open-source AI models for most queries.
• Can do basic Q&A, send emails, set alarms, and start a stopwatch. Adding more on a weekly basis.
It’s still rough, hardware is off-the-shelf dev boards, voice can glitch in noisy places, etc. But the core “ask wrist -> instant answer” loop feels surprisingly useful in real life.
If you’re the kind of person who hates breaking stride to grab your phone for trivia, or you’re into AI / wearables / low-distraction tech, I’d genuinely love brutal feedback:
• Does this solve an annoyance you actually have, or is wrist-voice too gimmicky?
• What queries would you actually use it for day-to-day?
• Hardware/form-factor thoughts? (Current one is a mediocre thick watch)
Waitlist for early prototypes is open, with a discount for the first batch if you’re interested in trying one: https://gethollow.com
(Founder here, solo-building this. Happy to dive into tech stack, why I chose certain trade-offs, battery math, privacy details, etc. in comments.)