The brief
Voice note apps have always been split. The capture half is solved — every phone has a recorder. The output half is not: you end up with a transcript that's literal and unstructured, which means you have to clean it up before it's actually useful. So most people stop using voice notes for anything but reminders.
The opening: in 2026, Apple shipped FoundationModels — an on-device LLM accessible from any iOS 26+ app — alongside an upgraded on-device SpeechAnalyzer. For the first time, you could take a voice note, transcribe it, and structure it into something useful — all without sending audio off the device. Asideo is a bet on that moment.
What shipped
Asideo 1.0 puts five hands-free triggers in front of the same recording flow, so capture happens before friction has a chance to talk you out of it:
- Back Tap — double or triple-tap the back of the phone
- Action Button — for iPhone 15 Pro and later
- "Hey Siri, Asideo" — App Intents integration
- Shortcut — for HomeKit, automations, or third-party triggers
- In-app — the obvious one, for when the phone is already in your hand
When you stop talking, the audio is transcribed on-device and handed to Apple Intelligence with a format instruction — diary entry, meeting summary, structured list — that you can swap per-note. The result is a clean, readable note you can share to the system share sheet or send to Notes, Mail, or anywhere else iOS reaches.
The technical bet
iOS 26.4 as the minimum is the load-bearing decision. It cuts off a chunk of the install base, but it's what unlocks the privacy story (nothing leaves the device) and the framing (Apple Intelligence, not a third-party LLM). The trade-off was made deliberately: a smaller addressable market for a meaningfully better product, with room to relax the OS floor as adoption catches up.
The pivot
Asideo didn't start this focused. An early version tried to do more — a soundboard sat alongside the capture flow, and the app's identity was split between "make sounds" and "make notes." It demoed fine. It just didn't have a point.
The turning point was admitting that the soundboard was a different app wearing the same icon. Cutting it back to a secondary feature — and rebuilding everything around a single job, talk and get a structured note — is what made the privacy story, the trigger lineup, and the name all line up. The growth wasn't adding features. It was having the discipline to remove them.
What's next
The 1.0 release is small on purpose. The next moves are about depth, not breadth: more structured output formats people can pick from, better defaults based on what you actually capture, and tightening the latency between "stop talking" and "have a useful note."