A few weeks ago I built and shipped a working app in under a week. No engineering team. No budget. No permission. Just me, at my kitchen table, with AI as my co-worker.

It's called JobTracker, and it's free. I built it for myself, honestly — I was applying for roles everywhere and drowning. Spreadsheets that fell out of date the moment I made them. A fog of "wait, did I follow up with that person?" Tasks slipping through the cracks at exactly the moment I couldn't afford to drop anything. Good tools existed, but they cost money, and I was trying to be careful financially. So I made the thing I wished existed and shared it out for free.

I've been genuinely moved by how many people have used it, and those who wrote to say it helped. A few asked how I built it. I'm happy to share — but the "how" turns out to be less interesting than what the "how" reveals.

The first version was a Claude Artifact. If you haven't built or used one before they are simple: you describe what you want, and in a few minutes you have a working app. It was fast and almost magical — perfect for one person. But it wasn't robust enough to hand to a friend who might rely on it during one of the most stressful seasons of their life. I wanted something real.

So I rebuilt it in Grok Build, which is made for deployable applications — the kind real people can depend on — at a fraction of what software like this used to cost. From there I kept going and built a second tool — AskHoot.ai, a decision-making app that's now live. (More on that one another day; the point here is the building itself.)

Here's the part I genuinely can't stop thinking about.

The barrier to creating a real digital tool is now roughly one $20-a-month subscription.