◇ Could this help me?
Yes. All three observations hold up independently of the podcast they came from, and the first is the one with teeth: when execution is cheap, the cost of a wrong direction is paid faster and in larger volume, which inverts the usual instinct to optimize for throughput. The alignment point is the least fashionable and probably the most accurate, since agreement is a human bottleneck that no amount of generation capacity touches, and it is telling that it is reported as unsolved at the company building the tools. General practice, with one honest note: taste is named as the thing that matters and not defined, which is the weak point of this genre of advice, so treat it as a prompt to work out what good looks like in your own domain rather than as guidance you can act on directly.
I want to pressure test a direction rather than execute faster on it. Take the thing I am currently building, described below, and argue the case that it is the wrong thing to build: what evidence would I expect to see if it were genuinely needed, what is the strongest alternative use of the same effort, and what am I assuming without having checked. Then tell me which people would need to actually agree for this to work, and where their understanding of it is most likely to differ from mine. Be direct rather than encouraging, and tell me what I should go find out before building more.