Brian Maucere

It's like it is the memory conundrum.

Right.

Which is no, you don't want and you've you've brought up many companies and research papers that are based in this.

It's such a big deal.

But it's like, no, we don't want all knowing memory.

It's just compute intensive.

Now, at some point where latency and compute drop to such a low dollar figure, we're talking about fractions upon fractions of a cent.

Sure, you might say to yourself, but what does it matter if my memory has two terabytes on it?

Or that it pulls in a half a gigabyte every single time it answers another question.

You and I might be sitting here three years from now and going, this is not a problem anymore.

They've solved for this.

But in today's world, tokens matter, costs matter, your premier models matter.

And so what I set out to do with my claw as a test was to build a faceless

CRM, essentially, that could fully do lead generation on its own by going out and connecting to the right tools and building that system.

And I have seen some hints of, hmm, that's interesting.

Because basically what it said was, you know, Greg will love this.

Like, you don't need the CRM.

You know, Carlos talked about that.

And I don't think that's right for everybody.

But in my particular case, the use case was a