Ahmed El-Kishky

Just all over the world, people are starting to think about using these models to, you know, sort of push beyond what humanity has currently discovered and known.

Maybe like in the future we'll have like, you know, papers where the first author is like an AI or something and it wins a Nobel Prize one day.

I mean, it's pretty basic, but yeah, Codex is really great internally because let's just say you want to make some change to, you know, our code base.

We can just like, you know, write something in there and just leave it and then come back and you have a pull request just ready for you or ask it to explain some part of the code that you don't understand.

Honestly, it's kind of shocking how useful it is and how much like you grow to rely on it.

Like we didn't have Codex, you know,

Like not too long ago, when I joined, everything was sort of done manually.

But now it's sort of such an accelerant.

And then obviously, like we use GPT-5 internally and just every time I have a question, I always ask it to, you know, explain things for me that I don't understand or, you know, work on some math for me.

So my personal stack is just GPT-5 and then Codex for anything related to code.

I don't know if they should be first or wait, but they should definitely learn the fundamentals.

One shouldn't rely on something as a crutch to not learn your fundamentals.

The fundamentals are always useful even beyond the immediate outcome.

It's very, you need to think long-term and long-term, you know, improving oneself involves getting the fundamentals.

But then once the fundamentals are sort of achieved and you've mastered them, it's an accelerant.

You should use it as a tool and then try to pick up something that's more abstract, more complex, or you can rely on this tool to sort of accelerate that.

Yeah, so I'm always pro-learning the fundamentals and I'll never say otherwise.

Thank you for having me.

It's been a pleasure, Corey and Grant.

I loved answering these questions.