Let me tell you how I got into AI. It wasn't some grand vision about the future of computing or a research paper that blew my mind. It was my kids. Specifically, it was hearing "I didn't know you wanted me to do that" for the four hundredth time and finally snapping. Not in an angry way — in a builder way. The way where your brain stops accepting the problem and starts architecting the solution. I'm sitting there thinking: we have four people in this house, everyone has things they need to do, and somehow the information just evaporates between the moment I say it and the moment it needs to happen. So I do what any self-respecting engineer would do. I build a todo app. For my entire family.
Now look, I know what you're thinking. A todo app. How original. The world definitely needs another one of those. But here's the thing — I wasn't building it for the world. I was building it for my household, which operates with all the complexity of a mid-size enterprise and none of the documentation. The kids need to know their chores. My wife and I need to coordinate who's picking up whom and when. There are groceries, school events, deadlines, and a hundred small things that fall through the cracks every single week. And the existing apps? They're either too simple to handle real family logistics or too complex for a fourteen-year-old to actually use. So I start building, and almost immediately I realize: I want this thing to be smart. I don't just want a list. I want something that can nudge, remind, and — this is the key part — eliminate the "I didn't know" excuse entirely.
That's when AI enters the picture. Not because I'm chasing a trend, but because the problem genuinely calls for it. I start playing with language models, figuring out how to make the app understand context. Not just "take out the trash" but knowing that it's Thursday, trash day is tomorrow, and this particular kid has ignored this task three weeks in a row. I want it to send the right reminder at the right time in the right tone. And suddenly I'm deep in prompt engineering, embeddings, notification logic, and user experience design — all for a family todo app. It's ridiculous and it's the most fun I've had building something in years. No sprint planning. No stakeholder reviews. No twelve-person meeting to decide the color of a button. Just me, a problem I actually care about, and the freedom to try whatever I want.
Here's what ten years in corporate never taught me that this one side project did: when you build for yourself, you learn faster than any training program, any certification, any carefully scoped Jira ticket could ever teach you. You make decisions in minutes that would take weeks in a corporate setting. You ship something broken, fix it at midnight, and ship again. You learn what AI can actually do — not from a whitepaper, but from trying to make it remind your teenager to unload the dishwasher before dinner. And the beautiful irony? That silly little family todo app is what set me on the path to building real AI systems. Everything I'm building now traces back to a dad who got tired of hearing "I didn't know." Did the app fix my kids? Absolutely not. They still "forget." The dishwasher is still full. The trash still sits there on Thursday night. Turns out, you can build the most intelligent reminder system on the planet and a teenager will find a way to ignore it. But that's not the point. The point is that I came out the other side knowing how to build with AI — really build, not just talk about it. The best projects don't start with a business plan. They start with a problem that's been driving you nuts long enough that you finally decide to do something about it. Even if the problem wins.