These 2 Tiny AI Models Run on Your Phone, Not the Cloud

Desert Ant Labs is back with two new tiny AI models that run entirely on your device: Who, which figures out who is speaking and when in any recording, and Eye, which tags, scores and sorts your photos without uploading a single image. In this video we look at how both models work, check the numbers, and test them with two demo apps: Who Said What for macOS, and an Eye photo sorter that runs in the browser. 🔗 Relevant Links Who: https://desertant.com/models/who/ Eye: https://desertant.com/models/eye/ Who Said What (macOS app, Who + Apple's speech transcriber): https://github.com/andrisgauracs/who-said-what Eye Photo Sorter (source): https://github.com/andrisgauracs/eye-photo-sorter Eye Photo Sorter (try it live in your browser): https://andrisgauracs.github.io/eye-photo-sorter/ ❤️ More about us Radically better observability stack: https://betterstack.com/ Written tutorials: https://betterstack.com/community/ Example projects: https://github.com/BetterStackHQ 📱 Socials Twitter: https://twitter.com/betterstackhq Instagram: https://www.instagram.com/betterstackhq/ TikTok: https://www.tiktok.com/@betterstack LinkedIn: https://www.linkedin.com/company/betterstack 📌 Chapters: 0:00 Desert Ant Labs is back 1:09 Desert Ant Labs in 40 seconds 1:51 Who: speaker diarization explained 2:16 950× faster than real time 2:39 Who vs pyannote: where it wins and loses 3:27 Who Said What: the demo app 3:50 Live demo: labelling a full podcast 5:28 Eye: a whole photo report in one call 6:08 Speed and confidence scores 6:59 Eye in the browser with LiteRT.js 7:48 Live demo: sorting 105 photos 10:35 Things to keep in mind 11:13 Final thoughts