Fair point, and that matches what we've seen too. The scaffold is genuinely fast now, but dispute flows and provider onboarding still need real design decisions, not just code. Where AI actually helps there is speeding up the iteration on those flows once you know what you're building, not skipping the thinking itself.

Here's a reply:

Completely agree, that's actually the bigger bottleneck in my experience. Code moves at whatever pace you push it, but Stripe Connect verification, payout test cycles, and getting a real provider to actually complete a test booking all run on someone else's...

Agreed, that split holds up well. The generic screens are basically free time now, but matching logic and payment edge cases are still where a product actually earns or loses trust, so that's exactly where the careful engineering has to go.

The other one that doesn't compress: app store review and processor KYC. Both run on someone else's clock and can each cost a week, regardless of how fast the build finishes.

Definitely, and those two combined can quietly double a launch timeline. App review can bounce for something as minor as a metadata mismatch, and KYC delays are basically out of anyone's hands once submitted. Best you can do is start both in parallel with the build instead of...

You are right!! AI has a great impact in coding...