[ruihanli]

I know two approaches. One is free-form rendering, like Anthropic's Artifacts. The other is to designate a category of tools whose output gets caught by the frontend and rendered in the chat using the corresponding components. I used the second one in several projects.

[mattsadowsky]

Ok, but what tech you are using second one approach? any open source, specific tech?

[ruihanli]

I built it by myself. If you want, I can share the code with you.

[mattsadowsky]

No, it is fine. Could you tell me only what langage you use to generate UI? HTML, JSON, MD? I what output you expect from AI Agent to generate UI component?