Build log
An iOS app for logging a project build to the minute and the dollar. Parts kept separate from tools. Logged hours kept separate from guesses. It runs, it logs hands-free, and it's in testing now — what's still open is which decisions become V1.
// no spam — just build updates as it's made.
// one person building this. you'll hear the real progress, good and bad.
This page sells an app that's real but not yet locked. It works — I log my own builds with it. What isn't finished is which calls become V1, and I'd rather decide that in the open with the people who'd use it than ship a finished-looking thing nobody pressure-tested. Everything below is working, designed, or honestly flagged.
Follow the buildThe real capture screen — clock on, work an area, switch by voice, clock off. Not a render, not live data: the actual flow you'd use in the garage, on repeat.
Not a feature list. The three jobs the app exists to do — and exactly how far along each one is.
Parts cost, cost-in-car, and total outlay tracked as three separate truths. Logged hours and estimated hours stored apart, always. The data model is built so a guess can't masquerade as a measurement.
"Suspension, start." The timer runs while your hands stay on the spanner, attributing every minute to the system you're actually working on — no stopping to tap a screen with greasy fingers.
One shareable image: your car as the hero, your parts cost and logged hours as the only numbers. A flex you can post, where every figure is literally what the timer recorded.
This was the one feature that could have sunk the app: voice logging in a loud garage — compressor running, music on, dirty hands, phone at arm's length. If that didn't hold up, the app was just button-tapping in gloves. So I built it first and tested it where it actually has to work: on my own phone, in my own workshop.
The bar was concrete and set in advance: at least 9 of 10 commands recognised across repeated garage runs, with every mishear caught by a spoken confirmation. It cleared it. Every command speaks back, so a mistake is caught by ear in the moment — not weeks later in the data.
And the app works fully by manual timer regardless — voice is the accelerator, never a dependency.
You can watch the spoken-confirmation flow in the looping demo above — every command speaks back.
Every build log on the site uses this exact data model. One's live now — read it, check the maths, that's the app's output today. More are on the way.
The app works and it's in testing now. One email when there's something real — the first public build, the day it hits the App Store, the calls I'm still deciding. If you'd use this, your feedback shapes what V1 becomes. No countdowns, no hype.
// iOS · in development · solo build.
// your email gets build updates and nothing else.