The export detail is the part most founders find out too late. Being able to download the code sounds like ownership, but if that code still needs the builder's runtime to work, what you own is a copy you cannot run anywhere else.

Your split between a quick internal tool and software the business depends on is the right way to frame the choice. For the first, a builder is often the sensible option, and the speed is real. For the second, the questions change to who can host it, who can fix it at two in the morning, and what happens to your data if pricing or terms change.

The six problems you found before writing any code, including the Stripe resend case, argue for a short review step before committing. A few hours of checking payment flows and data ownership is cheaper than a migration later.

I would suggest founders run the export test on day one: export, try to start it on a plain server, and see what breaks.