Rob Walling

I think there's a hypothesis that everyone's going to vibe code their own app.

And to Tracy's earlier point, it's like, no, we're not.

I used to write code for a living and I'm not gonna vibe code a bunch of crap.

I might do a little utility here, a little whatever, but the maintenance of that over time and the hosting, just the patches, like I don't want, and I get it, it could be Lambda, blah, blah, blah, whatever.

There's no way that my super smart PhD wife

There's no way she's going to vibe code a thing.

She's going to pay for software and most people will too.

And we're in these bubbles where we're all techies and we can all roll our own thing in a weekend.

That's fine.

Go do that.

But that's not Tracy.

You've seen how resistant that I've been to even writing, having any code that tiny seed maintains.

Cause it's like, I don't want to be a software shop.

I know that, you know, the headache and maintenance of that.

With that, I want to roll to our next topic, which is actually something Anar, you brought up already.

It's just a question.

There's no article here, but I had a question for the two of you.

And I'd like to start with Tracy again.

My question is, are the AI models that you are using today noticeably better than they were six, maybe 12 months ago?

We can kind of put a timeframe on it.