UluP SDK Studio is the visual editor for building automations inside UluP Spaces without writing code: drag a block, connect a trigger to an action, and from then on Spaces does the work on its own. Until now, the canvas was the only way to build them.
As of today, it isn't the only way anymore.
How it works
If you already use Ulupy to connect Claude to UluP Spaces, you can now also ask it to build an automation for you, just by describing it:
> "When I complete a task, save it as a reminder in Inbox."
Claude looks at the automations you already have, tries the new one in test mode (no real data touched, just a simulation), and then creates it. It's born always disabled, exactly like an automation built by hand in the editor: no AI can turn it on in your place. Flipping it on stays your own move, inside `/studio`.
Why this isn't "another AI bolted onto the app"
The obvious temptation would have been to have a model called on every real event, deciding what to do in real time. We considered it, but it has a concrete problem: someone has to pay for that, on every single event, forever.
This path is different. An automation born from a conversation is identical, byte for byte, to one born from the canvas: same declarative JSON, same validation, same execution engine. Only *who writes it* changes — you, in words, instead of you, dragging blocks — not *what runs afterward*. The cost of the model is paid by whoever is chatting at that moment, not by SDK Studio on every trigger that fires for years to come.
How to try it
1. Connect Claude to UluP Spaces via Ulupy, if you haven't already 2. In chat, describe what you want to happen and when 3. Review the automation (disabled) inside SDK Studio → `/studio` 4. Turn it on when you're ready
The canvas stays exactly as it was, for anyone who prefers building with blocks. It's now just one of two ways in.
---
*This update is live on UluP Spaces.*