Discovering AI felt like opening Pandora's box.

For fourteen years, music was the plan. I played in a couple of bands, and one of them, Happy Alone, started genuinely going somewhere. We were flown to LA to meet Geffen Records. There were deals on the table. Everything was pointing in the right direction.

Then Covid arrived and took the whole thing down. No gigs, no events, and one by one the label people who believed in us moved on. The momentum we'd spent years building disappeared in a few months.

What followed was years of trying things. New music projects. An e-commerce business. None of it stuck. Honestly: most of a decade of trying and failing, over and over.

Early 2025, my buddy Pete showed me Replit, an app that lets you build software from your phone. It was like opening Pandora's box, and the feeling was one I already knew: the first time I ever picked up a guitar. Whoa. I can do anything. Everything I made broke, and it was still the most exciting thing I'd touched in years. I thought: this is the future, and I need to learn to do it properly.

So I went all in. On a two-week family holiday I built from eight in the morning to two at night, every day. Antisocial levels of obsessed. My first app took those whole two weeks. It broke. I rebuilt it in two days. It broke again. I rebuilt it in one. By the end I could rebuild it in a couple of hours. That's when I understood: the skill compounds if you keep showing up.

I tried to turn it into a business, selling automations to companies. It mostly didn't work. By autumn I'd hit a wall: hours going in every day, nothing coming back. Then around Christmas some bad news landed, and instead of stopping me it pushed me deeper into the work. Nothing was happening, and nobody was coming to save me. So: lock in.

On New Year's Eve I made one promise to myself: post about this once a day, every day, and something will happen.

By March, something happened. First clients. First paying customers.

Today I run Tobin, an AI agency building systems for real businesses, and I've put 4,000+ hours into Claude and the tools around it. Everything I learn doing that work lands on this site as free, plain-English guides.

The reason I write them for complete beginners is simple: two years ago, that was me. If a musician who spent a decade failing at everything else can learn this, so can you.