Pulso’s Android release had never actually been shrunk: the app was shipping unminified at 87MB, R8 turned off since the day the project was scaffolded. Turning it on took the same app down to 3.75MB, the kind of number a Pixel 9 emulator won’t tell you is wrong until you try to hand someone a real file. Signing followed the same day, a real release keystore, so testers get a signed APK instead of a debug build with a warning label.
Build 24 went up to TestFlight the same day, the first upload since 21, carrying three builds’ worth of fixes in one go: sharing had been unswitchable because the name field it required had nowhere to be entered, the roadmap visibility toggle was decorative (the server ignored it and shipped the row regardless, confirmed by watching a probe rung ship while the switch read false), and the account panel on web became a hub of rows to match iOS instead of keeping its own layout.
The bigger fix was the landing page, not the app. Ten separate claims on Pulso’s own marketing page didn’t match what the app actually does: it named Chrono24 and WatchCharts as price sources when it uses neither, and the hero led with price research when the real first thing a new user sees is their own collection. Corrected against the actual feature set, one claim at a time, and the price numbers on the page are now labelled a rough idea rather than stated as fact.
What’s still open: the Friends flow, accepting a request and reading someone else’s collection, is built but unverified, because verifying it needs a second real account and that test hasn’t been run yet.
