@jmartisk do you think this is possible without too much work?

What would that mean exactly, that the response from the model is a Multi and we would forward the events from the Multi to the UI as they arrive, over a websocket?

Yes

Yeah that shouldn't be very hard. I can look into it, just not sure when, things are piling up now :)

Yeah, not a priority for sure, it would just be nice for demos

I can give it a shot.

I have something is working. However, it will only work when rag is disabled (and it's true by default).

Should we add a message / alert to let the user know when streaming chat is not enabled and why it's not enabled?
Would you think that a vaadin alert or a console.log is more appropriate for this?

Should we add a message / alert to let the user know when streaming chat is not enabled and why it's not enabled?

I think that's a good idea.

I am pretty sure that noone is going to look at the browser console though :)