I've tried dozens of to-do apps. Todoist, Things 3, OmniFocus, Reminders, Notion, plain text files, sticky notes on the monitor. I've gone through phases with most of them.

None had exactly what I wanted. Either too complex, or missing something small but critical. So I kept switching, and the switching itself became the problem: more time managing tools than managing work.

At some point I stopped looking for the right app and started thinking: what if I just built it myself?

I'm not a Swift developer. I had no real Xcode experience. But I was deep into exploring how AI was changing what's possible, and I decided to test that boundary directly. Could I build a real, working Mac app from scratch using Claude as my technical co-pilot?

"The game has changed. We no longer need to search endlessly for tools that fit our workflows. We can simply create them."

What I actually wanted

Before writing a single line of code, I wrote down exactly what I needed:

That's the whole list. Nothing fancy. Everything I actually used.

The Now / Nxt / Ltr system

The three lanes are the core of TodoNxt, and honestly the thing I'm most happy with.

NOW is what you're working on today. Not this week, not eventually. Today. It stays small on purpose.

NXT is your queue. Ready to go, waiting for NOW to clear. Having tasks staged here removes the cognitive load of figuring out what's next.

LTR is the parking lot. Ideas, someday-maybes, things you don't want to forget but shouldn't be thinking about right now. Out of sight, easy to pull up when the time comes.

The system comes from how I naturally think about prioritization. Most todo apps either give you one flat list or a complex project hierarchy. This sits in between. Just enough structure to think clearly.

Building it with Claude

I started with zero Swift knowledge and used Claude as my Xcode tutor, pair programmer, and debugger. The process was genuinely different from what I expected.

The hard parts weren't the features. Claude handled SwiftUI layout, drag-and-drop, UserDefaults persistence, menu bar integration, and even Sparkle for auto-updates. The hard parts were the things I didn't know to ask: macOS Gatekeeper behavior for apps distributed outside the App Store, code signing, building a universal binary for Apple Silicon and Intel.

Each time I hit something I didn't understand, I described what I was seeing and asked why. I learned more about macOS development in a few weeks of building than I would have in months of reading documentation.

The result is TodoNxt, a real signed app that installs on macOS 13+ and updates itself silently in the background.

What happened after I launched

I put it out at €2.99, one-time, no subscription. The pricing logic was simple: low enough that the decision is easy, high enough that it signals real software and not a hobby project.

12 sales so far. A small number, but each one is from someone who found the app, decided it was worth something, and paid for it. That still feels meaningful.

The traffic analytics told an interesting story too. Most visitors come directly. Word of mouth, shared links. A mention from the MacStories community sent a notable wave. Almost nothing from Google yet, which is why I'm writing this post.

What this means beyond TodoNxt

The broader point isn't about this specific app. It's that the barrier between "I wish something like this existed" and "I built the thing" has collapsed.

You still need to think clearly about what you want, understand your users, and ship something with care. AI doesn't replace that. But the technical implementation, the part that used to require years of platform-specific expertise, is now accessible to anyone willing to learn alongside the tools.

I built exactly the app I wanted, on the platform I wanted, in a fraction of the time it would have taken otherwise. And I learned a lot in the process.

If you want to If you want to try TodoNxt before building your own, it's €2.99, one-time, no subscription.

Get TodoNxt for €2.99 →