Guess I'm a bit less impressed seeing that for some of the more intelligent driven+action work -- splitting requests in the video -- they had to kick out to an anthropic model.
Haiku, to rewrite a sentence as two discreet commands.
I agree that it was notable that they delegated to an existing LLM, but I don't think it detracts much from the value proposition (not yet proven) of their demo.
That's fair, but it highlights how this would actually be used. It doesn't really seem like a competitor to other models but instead a way to make these real systems more enjoyable to deal with.