info.arnao.ai: start clean, put the sensors behind "Experimental mode"
_Proposal only. Nothing is built yet. 2026-10-04._
The idea in one line
A visitor lands on a clean portfolio card and gets no permission pop-ups. Their first tap plays your voice intro and brings up the controls. Camera, mic, gestures and shake all live behind one big Experimental mode button, and all of them start off.
Mockup (live site, three steps)
| 1 ยท Land | 2 ยท Tap anywhere | 3 ยท Tap "Experimental mode" |
|---|---|---|
 |  |  |
| Just the card, no buttons and no prompts. A faint hint says "tap anywhere for controls." | Your voice intro plays. The normal controls fade in, and a big oval Experimental mode button sits where the mic was. A small ๐ mutes your voice. | A short explainer with four switches, all off. The browser asks for permission only when someone flips a switch. |
Full-size frames: 1 ยท clean ยท 2 ยท controls ยท 3 ยท experimental ยท today's first screen, for comparison
What changes
- Clean is the default. The existing "lights-out" mode, where tapping the background hides the chrome, becomes the starting state. The "Let's go" card goes away.
- The first tap plays the intro. Phones won't play sound until the visitor taps, so the voice can't start on page load. The first tap anywhere plays your cloned-voice intro and reveals the controls in one move.
- The Experimental mode button contains:
- ๐๏ธ Talk to it: uses the mic
- โ Hand gestures: uses the front camera
- ๐๏ธ See-through background: uses the rear camera
- ๐ฑ Shake to meet Byron: uses motion sensors
It also has Turn all on and Done buttons. The choices are remembered on that device.
- Moved off the main screen. The ๐ and โ buttons leave the top bar, and the mic button leaves the bottom. Narration in your voice stays on, with a ๐ mute button.
Everything that asks for permission today (all of it moves behind the switches)
| Where | Today | After |
|---|---|---|
| "Let's go" tap | Rear camera, then mic (default mode is talk + listen) | No prompts. Default mode is "speaker only." |
| Any first tap on iPhone | Motion-permission prompt (armMotion on first pointerdown) | Only when the Shake switch is turned on |
| After the greeting | Live-voice session opens the mic automatically (rtMaybeStart) | Only when Talk is on |
| After the greeting | Spoken gesture invite plus a front-camera prompt (maybeInviteGesture) | Removed. Gestures are a switch now. |
| Returning visitors | Auto-resume whatever was granted before | Resume only what's switched on in Experimental mode |
Not changing
Cards, launch, details, search, filters, Stack, the Signal, and the narration clips stay exactly as they are.
Open questions for you
- Should the intro play on every visit or only the first? (Today, returning visitors get a short welcome-back clip.) Recommendation: keep it that way.
- On first tap, should your voice play automatically (as mocked up) or wait for a "โถ hear Byron" button? (Fable votes for the button.)
- Should the LAUNCH pill also hide in the clean view? In the mockup it overlaps the top of the card.
Self-critique โ Fable adversarial pass
Fable's main points, with how I'd handle each:
- Playing audio on the first tap is one surprise too many. Fable suggests revealing the controls silently and putting the intro behind a "โถ hear Byron" button. Your call. You asked for the intro to play the way it does today, so the mockup does that. The quick option is a ๐ mute that's always visible.
- "Experimental mode" sounds unreliable, and visitors won't find gestures. I kept your name. It's honest, and the pill is big. If you like, the explainer can lead with "try hand gestures".
- Prompting when a switch is flipped is the right pattern. Fable adds that a denied permission on iPhone never re-prompts. Adopted: a switch whose permission was denied shows "Allow in Settings โ Safari" and does not silently fail.
- A faint hint can be missed. Adopted: the hint pulses once and then stays faint.