Two public URLs, not just plans
In March, I wrote that I planned 28 apps and shipped 2.
I had many detailed and scored ideas, specs, analysis, priorities… A whole system for deciding what to build.
I’ve never been short of ideas. But now I’m getting them out there.
The useful thing I got from all that planning, though, was not the list. I think I finally got the hang of turning an idea into something real.
It started with MCPVault. I made it to learn MCP, but it also scratched a personal itch: I wanted Claude Desktop, ChatGPT, and whatever AI tool I was using to work with my Obsidian vault without copying context back and forth or messing up metadata.
It now has more than 1.5k GitHub stars and ~200k downloads. I did not expect that when I was just trying to make my own notes more useful.
This week, swipetris.com went live. And onedone.me went public too: a landing page and interest form for the free iPhone widget I’m still building. I shared OneDone on Reddit.
One is a free, no-ads falling-block game built for your thumb. The other is a tiny daily prompt to do one positive thing, then get on with your day.
They are very different things, but I wanted both of them to exist.
The plan is not the product
I still ideate and plan. Probably more than most people.
I now save the bigger models for figuring out the problem and turning a fuzzy idea into small tasks.
Terra, Sol, K3 on high. That is where I want the extra thinking.
Then I give execution to faster, cheaper models, DeepSeek V4 Flash, K3 or Terra Medium, sometimes even on low. They do a task, I skim the diff, test it, and move to the next one.
It sounds obvious written down. But I used to let every task become a conversation, then a redesign or a new abstraction for a problem nobody had. I ended up with so many ideas that I have an Obsidian vault just for that.
Rabbit holes are still fun. They are just not where things get published.
Publishing does not mean throwing something rough online. I still want it to work properly, be fast, and be beautiful.
By “beautiful”, I do not only mean rounded corners and nice type. I mean the engineering side too: a clean interaction, no weirdness when you use it on an actual phone. I still do actual human e2e testing, again and again, to see whether it makes sense. Small details that make a thing feel considered. I’ve written about that itch before.
A game for your thumb
Swipetris started as a simple question: why do so many mobile block games still feel like a console game squeezed behind a glass screen?
So I made one that is all gestures, where you drag to move, tap to rotate, and flick down to slam the piece.
It has one new board every day, with the same piece sequence for everyone. You can replay it as much as you want, but only your best score counts. A little friendly competition.
No ads. No account. No coins. No tracking. It works offline (PWA).
I built a game, and I usually don’t play games. But since you can mindlessly swipe up and down on social media while waiting for something, why not swipe some blocks around?
One positive thing a day. That’s it
OneDone is almost the opposite.
The widget is not live yet. The landing page is, with an interest form for when it lands.
The idea is one small daily action, delivered through an iPhone widget. Text someone you care about. Take a five-minute walk… Then you are done for the day.
If you miss a day, that’s fine. You just pick it up again tomorrow.
I wrote a little why page for it, because this part matters to me. The internet gave me most of what I know, connected me to people I would never have met, and made my life possible.
Somewhere along the way, though, so much of it became feeds, engagement loops, scorekeeping, and outrage. I cannot fix all that. But I can make small things that point the other way, and make them available for free.
The leaderboard makes sense in Swipetris. It would make no sense in OneDone.
Two URLs, not another folder
Swipetris is a tiny game that I made in one day. OneDone is still a landing page and waitlist while I finish the app. But they are online.
I still have lots of ideas. Probably too many. But two of them made it out of Obsidian this week.
That feels like a good start.