I built an iOS app for my sons who play club basketball. It’s designed to bridge the gap between practicing solo on the court and the pressure of playing in a game.
It uses mechanics and body movement, via visual detection through the cameras, to identify when you’re about to take a shot or make a pass.
It then pumps distracting noises (honks, cheers, jeers, etc.) into the headphones to try to distract you. It also runs a continuous 24-second clock so you build a natural sense of how much time is left before you need to get a shot off.
Building Basil (https://basilai.app), a privacy-first AI executive assistant / note taker. Everything is on-device: transcription, summarization, etc. So nothing is shared with any model provider.
Lately the interesting work has been less about raw transcription and more about making the output actually useful: recurring workflows, follow-ups, and personalized summaries that fit how someone runs their day. We are also pushing more toward reliable agent automation instead of one-off chat. And dealing with all the iOS limitations on background work.
Building a voice recorder/transcription/summarization app where everything runs on-device. No cloud, no uploads, no servers processing your audio.
Started with a web version using WebAssembly Whisper to validate the idea. Worked well enough to prove demand, but browser sandbox limits (no background recording, inconsistent model performance) pushed me to go native.
Rebuilt as an iOS/macOS app with CoreML. Apple Silicon handles Whisper-class models locally without issues, but its confined to recent devices.
I'm working on e2ee sync between web and mobile. I want to build a simplenotes but for audio transcription.
My experience share only supporting the latest OS:
I have launched apps focused on a new feature in the latest OS and regretted it. The # of people who have the latest OS is much smaller than the full install base for much longer than I thought. As a result, my marketing conversion was unnaturally low - people who liked the app idea but couldn't install because they had the wrong OS. This causes two problems: potential users I activated but couldn't convert and this signal gets internalized by the App Store, pushing down future impressions.
Now I always have a fallback implementation of the feature so I can target the prior OS. Both Mac and iOS.
As a side project, I just launched a privacy-first web-based meeting transcriber (https://basilai.app/app). Everything runs entirely in your browser — both the transcription and AI summarization — so no audio or text ever leaves your device.
I'm using the browser built in transcription service plus downloading a model and running it via webgpu. No login. At the end of your meeting, you get a zip file with the audio, transcript and summary.