Airspace

Apple had published its spatial design guidance for Vision Pro, and I used it as the starting point: frosted glass panels with real depth, soft rounded shapes, and content that floats in front of an environment instead of sitting in a box. But our users were on laptops, not headsets, so anything they already knew from the web stayed. Buttons look like buttons, menus sit where menus usually sit, and you can do everything with a mouse.

The layout of the room came out of that. Everything has a fixed place: the people in the middle, what you can open on the left, who you're with on the right, and one row of controls along the bottom. Learn it once and you know every space built on it.

Anyone arriving from a link gets one screen and one question: their name. There's no account to create, no app to install and no camera setup wizard. Type your name, press the arrow, and you're in the room.

The butterfly in the middle doubles as the loader, so the few seconds it takes to build the space read as part of the welcome instead of a spinner. And the screen carries the client's brand, not Mobeus's. Someone joining a United Therapeutics session should feel like they've arrived somewhere United Therapeutics made.

The usual move on a video call is to shrink everyone to thumbnails the moment somebody shares a screen. It's efficient, and it turns a conversation into a presentation. In a space that's meant to feel like being in the same room, that was the one thing to avoid.

So when a video, a device guide or a quiz question opens, it takes the centre and the two people swing out to either side, angled toward it like two people standing at a counter. They stay full size and you can still see their faces react. Close the content and they swing back.

The two people in the room need different things. The customer needs the product. The rep needs to know who they're talking to. So the context panel on the right is for the host. In the Verizon store it pulls the customer's Salesforce record: case history, current device, plan and what they came in wanting. In the training room it shows the trainee's module and how far through it they are.

The AI suggestions live there too. When the conversation turns up something worth recording, like a customer settling on an iPhone 15 Pro Max, a small card offers to update the CRM. One click, and nobody has to stop talking. A Host View and Customer View switch at the top lets the rep check exactly what the customer can see.

The room takes voice, hands in the air, and a mouse. The risk with voice and gesture is that nobody knows what they're allowed to say or do, and they won't guess. So the prompt bar at the top lists the words that work right now, like "details", "next", "previous", "go back" and "close", and the list changes with what's on screen.

Every one of those commands is also a button. Hand tracking is a toggle in the dock, not a requirement, and nothing in the room can only be reached by waving at it. The gestures are the part people remember from the demo. The buttons are why the demo works for the person who has never tried it.