AI just replaced your UI layer.
What if AI didn’t just respond with text
but generated the actual interface for you?
That’s what Thesys C1 does.
You send a natural language prompt.
It streams back live, interactive components.
Not another chat wrapper.
Not a markdown block.
Not a code snippet you still have to wire up.
A rendered interface, already working.
Ask it “show me monthly revenue trends” and you get a live chart.
Ask it “list users from the last 30 days” and you get an interactive table.
The glue code just disappears.
They’re also open-sourcing the standard.
OpenUI is their proposed open standard for Generative UI.
Think of it like HTML, but for AI-generated interfaces.
The spec is public.
Anyone can build on it.
The switch is surprisingly low friction.
The API is OpenAI-compatible.
Two steps: change the base URL, swap your markdown renderer.
That’s it.
Why this caught my attention.
I’ve spent 15 years writing glue between backend logic and frontend UI.
This is the first thing I’ve seen that directly targets that seam.
Not by abstracting it.
By eliminating it.
My question for you.
Has anyone shipped something with C1 or OpenUI in production?
Are there tools you’d compare this to?
What breaks at scale?
Leave a Reply