it still takes the agents time to build correct software, in my experience

The adage holds, you're just measuring at the wrong timescale. Fast and cheap is absolutely happening right now. But "correct" in software was never about whether the code runs today it's about whether the system is still understandable, maintainable, and predictable six months from now when requirements change and the person who wrote it is gone. LLMs are remarkably good at local correctness. A function works. A test passes. But software systems fail at the joints the places where assumptions from different parts of the codebase meet. That's where models break down, and that's precisely where LLMs are weakest, because they have no persistent model of your system's evolution over time. The triangle didn't disappear. It just got a time dimension added to it. Fast, cheap, correct right now maybe. Fast, cheap, correct over time still pick two.

It's one of those glib statements that I never really believed.

If you got at something with the wrong mental model you are always going to be pushing a bubble around under the rug and you're going to feel tied up by constraints, find "correct" always elusive not matter how much time you take.

If you go at something with the right mental modal it often falls into place.

The "fast vs cheap" dilemma in a well-run operation is that once you have got efficient development under control (cheap for real) you can spend money to accelerate the schedule. "Efficient" and "under control" pretty much require correct. On the other hand I'd expect the average person using this slogan glibly is working on a project that will be late, expensive, and terribly incorrect.

Still true in the age of LLMs. They are fast and relatively cheap, but correct is another story.