Very few companies have what would be considered a proper large language model with a dataset that is theirs and useful in the right way it’s a lot of work – OpenAI didn’t build it overnight

I think a lot of people think they have to have it. It’s come from the top. It’s you know, like a first order priority or close to it. So, it’s almost difficult to talk about it without like asking what have you really done? Yeah. And I think maybe to clarify, people mean different things when they say they’re building their own LLM, right?

I just had a meeting with a prospect yesterday. The upper echelons of management believe that they need to train their own LLM that’s going to encompass all of their internal knowledge, customer information, policies, etc. I think that the type of organizational action that’s triggered when FOMO—the fear of missing out—is the trigger is quite different than when you actually have an active problem to solve.

I think people are actually in this mode of accumulating assets. It’s oh, I want my own LLM so that I’m not missing out. But they’re not really thinking about what problems they actually want to solve with it. It’s a fairly interesting part of the adoption curve.