AERA · PERSONAL AI STUDIO
Say what you want. Get a module that's yours.
AERA is a personal AI workbench. You don't learn a stack of tools — you say, in plain language, what you want to do, and it builds a module around that, running in a deterministic workbench you can return to. Right now it's an iOS local prototype, still in development.
What it is
Say, in plain language, what you want to do, and AERA generates a personal module that's yours. The module then keeps running as a deterministic workbench — use it today, pick it up again tomorrow. What you see and what it can do are declared up front, with no silent improvisation.
Not a fixed-template app
Every module is shaped by your own description, not picked from a set of templates.
Not a multi-agent console
No swarm of AI agents running around for you — just one quiet workbench you can read and control.
Sample modules
Two everyday scenarios, as a start
The first two sample modules already generate and run locally. A module is described in natural language, written as a structured schema, then drawn by a deterministic renderer — what you see is exactly what was declared.
Learn English
Runs locally
schemadeterministic renderer
Ten minutes a day: vocabulary cards, dictation, and reading practice with feedback. Describe your learning goal and AERA builds the whole module.
Fitness log
Runs locally
schemadeterministic renderer
Workout plans, sets and reps, progress trends. Tell it your training routine and get a module shaped around it.
Themes
Three themes, a quiet studio
AERA is a calm personal studio — every module can wear one of three themes, restrained and unhurried.
Clear
Bright and airy, for daytime
Slate
Muted stone grey, for focus
Canvas
Warm paper white, close to handwriting
VERIFIED BASELINE
What's verified today
Everything below is real: exercised by automated tests in the Simulator, with screenshots reviewed by hand. It shows the prototype works as designed — not that it's released or shipped.
77 / 77
Automated tests all green
20 UI + 57 unit tests, 0 failures, 0 skipped.
First sample modules
'Learn English' and 'Fitness log' generate and run locally.
Module switching + menu/card workspace
Switch modules and work in the menu and card-based workspace.
Checklist state, persisted on-device
Checked rows are saved on this device and survive relaunch.
Three themes: Clear / Slate / Canvas
Switch anytime; the interface and cards redraw per theme.
Bilingual interface — 中文 / English
Interface copy is available in Chinese and English.
Accessibility baseline
Dynamic Type, VoiceOver, and Reduce Motion verified in the Simulator.
This is Simulator automation and screenshot review — not a real-device walkthrough and not App Store review. A manual VoiceOver pass on a real iPhone is still on the to-do list.
Safety boundary
Generated structure, not a black box
Right now, generation happens entirely on-device: no network, no remote AI service. AERA outputs a structured module — screens, state, and actions all declared — which is validated before a deterministic renderer draws it. There's no implicit improvisation: what you see is what was declared, and when it can't do something, it says so plainly.
A module = schema + renderer
{ screen: 'wordCard', state: { deck, streak }, action: 'flip' }Local, deterministic generation
Module generation happens entirely on-device today, with no network requests.
Honest when it can't
When it can't do what you asked, it returns 'not available' plainly — it never invents a result.
Structured and validated
A module is described as a structured contract and validated before rendering; unsupported formats or capabilities are marked honestly.
Generated code is never executed
No AI-generated SwiftUI, HTML, or JS is ever executed — only declared structure is rendered.
NOW · NEXT
What works now, and what comes next
This keeps a strict line between what already works in the local prototype and what is planned but not live yet. AERA is a real local prototype — the following are not live:
Now (in the local prototype)
LIVEiOS local prototype
A personal workbench that runs on your device.
Two sample modules
Learn English and Fitness log, generate and run today.
Local deterministic generation
Module generation and rendering don't depend on the network.
On-device persistence
Checklist state is saved on this device.
Next (not live yet)
PLANNEDReal AI generation on a server
Server-side AI module generation — not live yet.
Accounts / cloud sync
Sign-in and sync across devices — not live yet.
Weather preview
A planned lightweight weather preview, likely using WeatherKit with your permission for location, showing the latest available data and when it was updated — not live yet.
HealthKit on a real device
Real health-data access, requiring real-device validation — not live yet.
Web app / WeChat mini-program
Cross-platform experiences — not live yet.
TestFlight / App Store
Public testing and release — not open yet.
No order or date is promised for any of these; each will be updated honestly here and on the product page.
Current status
An honest statement
None of these are live yet. AERA is a real local prototype, but the following remain in development — no fake links, no placeholder QR codes here.
Download and preview links open once the product ships — no fake links here for now.
- App StoreNot available
- Mini-programNot available
- Web previewNot available
- Accounts / syncNot available
- Real AI backendNot available