A video meeting is very good at showing faces. It is less good at showing what somebody means by "this bit here." Screen sharing helps until several people want to inspect the same object from different directions. Then the meeting turns into a negotiation over who controls the camera.
The Spatial demonstration at CES makes me think about that problem. Its avatars and shared objects suggest a meeting arranged around a scene. The interesting part is the possibility of having a common reference that nobody has to describe through a rectangle. The CES demonstration shared by TechCrunch is a useful starting point, although a demonstration alone cannot tell us how comfortable a full working session would be.
Consider a small exhibition stand. One person cares about the visitor's route, another about the height of a screen, another about the equipment concealed behind it. A single perspective view makes those conversations compete. In a shared scene, they could look at different aspects while still pointing to the same structure.
That requires more than putting three avatars beside a model. Everyone needs to agree on scale and orientation. If I say "the panel on the left," left could mean my left, your left, or the model's left. A persistent marker attached to the panel is a better piece of communication than an increasingly elaborate verbal description.
There is also a difference between pointing and editing. Moving a shared object without warning can disorient everyone looking at it. An interface needs to show who is manipulating it and distinguish a private inspection view from a change to the common scene. These are mundane interaction rules, but they determine whether the room feels cooperative or chaotic.
I would judge a spatial meeting by the moment after somebody asks a question. Can the group bring the relevant material into view quickly? Can a participant inspect a detail without interrupting the conversation? Can someone without a headset still contribute? Those tests matter more to me than whether the avatars have convincing jackets.
The format will not help every meeting. Reading a paragraph, editing a spreadsheet, or making a short decision can be easier on a conventional display. A headset adds setup and physical friction. Spatial organization earns that cost when the arrangement of things carries useful information.
There is a smaller opportunity here too: preserving the working context. A room could retain the model, the annotations and the unresolved questions between sessions. Returning would feel like walking back to a workbench, rather than searching for the right attachment in a chat history.
I like the direction because it gives remote collaboration a concrete design problem. Build a place where a group can inspect something together, make a decision, and leave a useful trace. The virtual meeting room can then justify its existence through the work that happens inside it.