Karl Yeh

But there's, I keep coming back to that, that article by Jaya Gupta where he was saying it's like decision traces and how things are actually decided is not through text.

It's not through a Slack thread.

It's not through that.

It comes through all the emails, what's in your head, Slack messages, just a bunch of things all over the place to have decisions.

And an agent can actually capture that and understand that and think through that.

And problem solved until it's done.

Not Ralph Wiggins style, but actually true.

Not just bash your head into it until it's complete, but taking the learnings continue.

There's a good post by... Who's the guy who just left OpenAI?

One of the researchers who started his own company, Jerry something.

He wrote a post saying how you can't get AGI until...

that intelligence component is there that it can self problem solve, not just loop solve.

Like there's a good post on that.

I thought it was, I thought it was a neat thing.

So yeah.

And he mentioned open clock can't do that.

Like some people will tell you it can, but it can't because it still requires you to prompt it into it.

I'll, I'll, you know what, I'll check the, once I go back to my virtual, I want to see what they actually did with it, which will be interesting, which will be really interesting.

So yeah, I'll report back tomorrow because I got to run, but like, yeah, yeah, no, I'll, it'll be interesting what those 10,000 did.

Cause I didn't give them any specific outside of go, you know, go into each molt and,