DeskRhythm beta, My Next localization, and a vibe-coding reality check

Three weeks since the last entry. That one ended with "DeskRhythm beta link out this week if Fastlane cooperates." Fastlane cooperated.

DeskRhythm is in public beta

DeskRhythm's native iOS app hit public TestFlight on June 26. Anyone with an iPhone can join the beta — no invite, no waitlist.

Since the week-24 entry, the main additions were Lock Screen and Dynamic Island countdowns that persist when the app is backgrounded, Rooms that carry their own work/break cadence and switch by location, walking-pad sessions logged to Apple Health, and iCloud sync across devices. Both languages throughout.

Most of the work between "beta-ready" and "public beta" was last-mile: Fastlane automation for TestFlight uploads, metadata and screenshots in both locales, the CloudKit production schema deploy, and a bug sweep that turned up more Live Activity lifecycle edge cases — the kind of thing that only surfaces across real workdays.

The product page now leads with the native app — new iOS screenshots, a beta signup link alongside the existing web link, and a rewritten story section that reflects where DeskRhythm actually lives now.

My Next goes from two languages to twenty

The My Next family — List, Note, and To Do — has been Swedish and English since launch. Now there are eighteen more: Arabic, Chinese (Simplified and Traditional), Danish, Dutch, Finnish, French, German, Hindi, Italian, Japanese, Korean, Norwegian, Polish, Portuguese, Russian, Spanish, Thai, and Turkish.

All eighteen are machine-translated. I can't justify the cost of human translators for apps at this scale — and honestly, auto-translation has gotten good enough that I'm willing to bet on it. If someone in Tokyo or São Paulo opens My Next List and the strings feel off, that's a real cost. But shipping English-only to those same users is a bigger one. The bet is that decent coverage in twenty languages beats perfect coverage in two.

The translations ship with the next update across all three apps. If quality problems surface, I'll fix them — but I'd rather learn that from real usage than wait for a budget that doesn't exist.

Vibe coding vs. structured workflows

I started a new project recently. It's still early — not even public on the website yet — and I kicked it off in Agent Dashboard with the usual PRD-driven workflow.

Then I wanted to move faster. I switched to vibe coding: just prompting Claude directly, one feature at a time, skipping the structured pipeline. Features went in quickly. The POC moved forward faster than it would have through Agent Dashboard's review gates and companion specs.

Then Fable became available again for a few days, so I pointed it at the codebase for a review. It found a fair amount of problems.

The data access layer had drifted — broken in ways that weren't immediately visible. There were security gaps, the kind that don't surface when you're the only user testing happy paths. Test coverage was thin. The project had accumulated the kind of structural debt that happens when each prompt only sees the feature it's building, not the system it's building into.

None of this was surprising in hindsight. Agent Dashboard's workflow exists precisely to catch these things: the review gates, the security audits, the companion specs that force you to think about the system before you start typing. Skipping them saved time on each individual feature but created work that now needs to be unwound.

The takeaway is straightforward. Vibe coding is fast for getting a proof of concept off the ground. But for anything that needs to hold up — correct data handling, real security posture, adequate test coverage — the structured workflow produces better outcomes, even though each step takes longer. The overhead isn't overhead. It's the work.

What's next

DeskRhythm beta feedback — I'm using the app daily and watching for rough edges that only show up over real workdays. If the beta holds, App Store submission follows. And the new project goes back through Agent Dashboard to clean up what the vibe-coding sprint left behind.