Corey Knowles

Copy again.

That's actually what Cloud Code Coursers does.

It's running a loop, basically.

But of course, it's a lot better that the program is doing it rather than the human doing it.

But if you think about knowledge work,

That's where knowledge work is, is basically from what I'm doing, probably what you're doing is we copy stuff from Google Docs into ChatGPT, we generate it, we paste it back, we look at it, like, it's not quite right, let's copy it back again, paste it back.

So there is, it's almost like if,

you are almost the API, if that makes sense.

It's like you are doing the grunt work of schlepping the data back and forth between your work systems, whether it's Slack, Google DOS, Google Drive, whatever, and ChatGPT.

And if you think about it from sort of a coding perspective, that is why Cloud Code and Cursor are so powerful, is you glue the LM directly to the task to be done, and it just goes.

Instead of you having to copy, paste these back and forth.

Yeah, and reformat.

I mean, that's what's happening to me today.

Every time I have to copy and subtract UBT, I have to reformat.

I have to copy as plain text.

I have to press tabs to fix it.

It's like unbelievable.

And so I think a lot of CIOs now are starting to realize that, yes, you could have this AI mandate, but honestly, just buying ChatGPT is almost so primitive.

You have to go beyond that.

You have to figure out, how does the LLM actually... How can I glue it to the task, basically, and have it do the task, the entire task from start to finish?