After the night of real-time collaboration, this was the session where the pieces really started clicking together. UluP Spatial stops being just "look at and move the nodes" and starts feeling like an actual working tool, with all its parts connected — literally.
Nodes connect to each other
It was the last thing plainly missing compared to the normal Spaces canvas: now you can tap a node, choose "connect to another node," and tap the second one — the line appears instantly and follows both nodes even if you move them afterward. Same connections you see on the website, same table, no copy. Deleting one from inside the headset isn't there yet — for now you remove it from Spaces, that'll come later.
A real focus mode, for every task
Before, a node just had a task list with a checkbox. Now tapping a task's text opens a dedicated view: you can write notes (they stay put, nothing gets lost), start a pomodoro-style timer, and see attachments already uploaded to that task — including audio recordings, which you can play with a tap.
The detail I care about most: the timer no longer resets if you switch nodes or close the panel. It lives one level above the panel itself, so it keeps running in the background — and a small indicator stays visible in the scene while it's active, so you never lose track of it by accident.
Attachments are view-only for now: uploading new ones from inside an immersive session would need a file picker that simply doesn't exist in VR/AR — a deliberate choice, not a forgotten limitation.
A small sign when the AI is at work
This is the part I like the most. If I ask Claude — from my phone, my computer, anywhere — to add a node or a task to the project using the same MCP tools I've been using on Spaces for months, and I happen to be in the headset at that moment, a small lilac glow appears on the node it just touched: a quiet sign that says "the AI is working here," right in the same space I'm standing in.
It's not (yet) a voice conversation with the AI from inside the headset — that would need a backend paying for every single request out of pocket, and it's not sustainable right now. But the real-time reflection of what the AI does elsewhere, yes: that's free, and it's already live.
The rest: real icons, not emoji
Small thing, but it was needed: every emoji scattered across the interface (the connection link, the keyboard, the notes pencil, the attachment icons) has been replaced with hand-drawn icons, in the same visual language as everything else — no more risk of a symbol looking off or not rendering well depending on the headset's browser.
What's still missing
Three things, in order of how much they bug me: deleting a connection from inside Spatial, uploading real attachments from the headset, and the actual release on the Meta Horizon Store — the last one is now just a matter of time, not decisions still to make.