First, can we hear about how this work started and what internal conversations or business cases initiated this project? Ari, may I ask you?

Yes, sure.

I think my thoughts about Black Box of LLM kind of started from the beginning.

I personally, I mean, this is a bit of an anecdote that I generally, I never really, I didn't want to believe that large language models, the next token prediction machines can actually do anything useful.

So I was to some degree wanted to look inside the black box myself to get more confident around it, because I'm not, I'm one of these people that doesn't really be like, you know, believes in hypes.

So I need to see it myself.

And I think I kept thinking about it in a particular, you know, we talk a little bit in our papers.

that