The first version was ugly. It also got the first cheque.

James: You've told me you were almost too embarrassed to show the first build. And that build is the one that got you paid. Reconcile that for me.

Hannah: The part the customer cared about — getting their stock counts right — worked perfectly. Everything around it was held together with tape. No settings page. No nice onboarding. I logged in as them to fix things manually for the first month.

James: So you spent your effort on the one job they were hiring you for.

Hannah: Exactly. They didn't buy a settings page. They bought never doing a manual stock count again. So that one path was immaculate and everything else looked like a student project. If I'd spread the polish evenly, the thing they were paying for would have been mediocre.

James: What did you deliberately leave broken?

Hannah: Billing was me sending an invoice from my personal email. Reporting was a screenshot I made by hand on a Friday. Multi-user was "share the password." All of that was fine for one customer who wanted one outcome.

James: When did the duct tape have to go?

Hannah: Around customer four, when I was the bottleneck. That's the signal — not "this is ugly," but "I personally cannot do this again." Until then, the manual back-end was the cheapest way to keep the promise while we learned what the product actually needed to be.