BUILD LOG // PAYDOWN
BLACK BOX
Building PayDown in public: a BNPL payoff app built with Claude, from first prompt to TestFlight.
TRANSMITTING · 16 ENTRIES · PAGE 1 OF 2
This blog is part of the plan
If a weekly public artifact is the rule, the artifact needs somewhere to live.
If a weekly public artifact is the rule, the artifact needs somewhere to live. So this site got built in a day, the same way everything else in this project gets built — me steering, Claude building.
READ →Marketing, accounts, and keeping financial data safe
Before going further I wanted three things nailed down: the marketing plan, whether PayDown needs accounts like Rocket Money has, and what "secure" actually…
Before going further I wanted three things nailed down: the marketing plan, whether PayDown needs accounts like Rocket Money has, and what "secure" actually means for stored financial data.
READ →37 apps since 2008. Why no hit?
PayDown is the 37th mobile app I've built since 2008.
PayDown is the 37th mobile app I've built since 2008. My first — a vacation countdown widget — sold about 2,000 copies and is still on the Play Store. None of the thirty-seven ever really took off, and mid-build I finally asked the question out loud: why can't I make an app that makes money?
READ →Taking PayDown to the cloud (carefully)
I wanted my numbers visible from a website too — which meant deciding what "cloud" means for an app whose best privacy feature is not having your data.
I wanted my numbers visible from a website too — which meant deciding what "cloud" means for an app whose best privacy feature is not having your data.
READ →The totals didn’t match Affirm
PayDown told me I owed $9,530.85.
PayDown told me I owed $9,530.85. The Affirm app said $9,153.92. For an app whose entire promise is one trustworthy number, a $377 disagreement is disqualifying.
READ →Why was "extra per month" capped at $500?
Playing with the strategy screen, I hit the ceiling on the "extra per month" slider: $500.
Playing with the strategy screen, I hit the ceiling on the "extra per month" slider: $500. Why $500? No reason. A slider needed a maximum and $500 got picked out of thin air — a developer default that silently became product policy.
READ →Payment history was missing
I imported a loan I'd been paying on for months, and the payments I'd already made simply didn't exist — no history, and no way to add them.
I imported a loan I'd been paying on for months, and the payments I'd already made simply didn't exist — no history, and no way to add them.
READ →Swipe, notes, and the Lead Bank problem
Three more findings from daily use, all shipped in build 4.
Three more findings from daily use, all shipped in build 4.
READ →The 3,361% APR
I pasted a real statement into the import screen and got back an account with a 3,361% APR and a balance of $138.51 on a much larger loan.
I pasted a real statement into the import screen and got back an account with a 3,361% APR and a balance of $138.51 on a much larger loan. Garbage in every field.
READ →Who do I owe?
First real session with the app on my phone surfaced two problems in ten minutes.
First real session with the app on my phone surfaced two problems in ten minutes.
READ →From empty repo to TestFlight
Two working days after the first commit, I told Claude to get it on my phone.
Two working days after the first commit, I told Claude to get it on my phone. Apple made that... educational:
READ →Calendar math is hard (so the engine has golden tests)
The heart of PayDown is a pure TypeScript payoff engine with zero app dependencies — the same math runs on the phone and, later, on the web.
The heart of PayDown is a pure TypeScript payoff engine with zero app dependencies — the same math runs on the phone and, later, on the web.
READ →