I’ve recently carved out a small window of time each week to return to hands-on coding, with the help of AI tools. This post is a reflection on what it means to re-engage with a long-standing side project — MealDeck — while balancing the demands of work, life, and everything in between. It’s also the beginning of an experiment: how far can I get with just eight focused hours a week?

Every software engineer has a graveyard of side projects — half-built apps, frameworks with pun-laden names, README files filled with bold ambition and little else. MealDeck is mine. Unlike the others, it keeps finding its way back to the top of the pile.

MealDeck is a simple idea: turn the drudgery of meal planning into something closer to a game. Make it collaborative. Make it fun. I’ve been circling this idea for the better part of three years — sometimes actively building, other times just staring at it from across the room while life took over.

And life, of course, is busy. I’m an engineering manager, a husband, someone who tries to move their body occasionally and put the bins out on time. Carving out real, focused time for side projects is hard — and not in the romantic, ‘I’m-just-too-passionate’ way, but in the actual, ‘I’ve-done-two-loads-of-laundry-and-I’m-knackered’ way.

But something’s shifted recently. The rise of AI coding assistants has opened up a new kind of possibility — not just in what I can build, but how I can build it, and more importantly, when. With a bit of planning and a few compromises, I’ve managed to claw back around eight hours a week. That’s not a lot. But it could be enough — at least, if I’m smart about how I use it.

This is, in a way, an experiment: can a time-starved engineer with more management experience than current stack familiarity build something real, meaningful, and shippable in just eight hours a week with AI’s help?

I’ll be using Claude as my main coding assistant, and ChatGPT to shape ideas and narrow down decisions before they reach the editor. Between them, I’m hoping to stay out of the weeds and avoid the trap of tinkering for its own sake. I still bring something to the table — over a decade of SWE experience, a reasonably sharp eye for product decisions, and a relentless sense that this idea is worth building.

What I find exciting is not just the chance to make progress on MealDeck, but the opportunity to test the limits of what focused, AI-assisted work can achieve. We talk a lot about velocity in software. What happens when the velocity isn’t in the hours, but in the leverage?

Let’s see. Eight hours a week. A strong idea. A few good tools. And a bit of persistence.