About

I'm Mike. By day I have a normal 9 to 5; by night, once the kids are asleep, I build things with AI.

This site is where that work lives: a timeline of projects I've shipped, sold, paused or quietly let go, plus a blog covering what worked, what didn't and what I'd do differently next time.

What I do by day

I've spent ten years in logistics, and worked my way up from the warehouse floor to operations, and now into business solutions, building and coding the systems that keep my current employer running more efficiently. That path matters more than it sounds: before AI, it meant I understood the whole supply chain from the inside, picking, put away, dispatch, planning, the systems that connect them and the ones that quietly don't. Now, with AI in the mix, I can see exactly where it fits and where it doesn't. Not theoretical use cases from a slide deck, but the actual bottlenecks a warehouse or ops team hits every day.

That combination, real operational experience plus building AI-assisted software myself, is what I bring to the consultancy work below.

How I got here

It started back in 2022. I was scrolling Instagram and Gary Vaynerchuk posted something plugging ChatGPT. I googled it, signed up, and then didn't open it again for weeks.

I'd been messing about with websites since school, simple HTML pages in Microsoft FrontPage, then PHP and MySQL once I found Tizag, building in Dreamweaver with XAMPP running locally. I was always the guy who built things and never shipped them. Mostly fear: people saying it was a rubbish idea, no feedback at all, or just failing outright. Some of it I couldn't finish anyway, because I didn't understand PHP well enough to build anything complex.

At the time my brother and I ran an online beer business. To help SEO we blogged about the products and the industry every week or two, until I couldn't be bothered writing them myself and found a copywriter on PeoplePerHour to do it for about a tenner a post. Next time a post was due, instead of messaging him, I thought I'd see what ChatGPT could do.

My first conversations with it were nothing serious: "do you feel love", a workout plan, 500 words on alcohol free beer to test its writing. Then I found a Chrome extension called AIPRM that injected all sorts of prompt templates into ChatGPT. One of them would "rewrite" a competitor's blog post from a URL. It wasn't actually reading the page, web browsing wasn't a thing yet, it was just working from the slug in the URL, but it did a decent enough job that I got properly curious.

So I asked it to build something in PHP, and it wrote code. Basic stuff, but I was buzzing. I could just ask ChatGPT to write code for me.

Becoming the "is it possible" guy

My own ideas mostly stayed ideas, but it meant I could build things at work. I became the person colleagues would ask "is it possible to do X" or "can you build Y", and I said yes to everything, then asked ChatGPT to write it in PHP before presenting it to the team.

A company we worked with built everything in Laravel, which I'd never touched. I asked ChatGPT to teach me, and about a week later I had a Laravel project running, still fuzzy on controllers, factories, models, middleware and auth, but asking ChatGPT to explain each one as I went. To an actual developer that's basic stuff. To me it was brand new, and it meant I didn't have to post on Stack Overflow and get torn apart for asking a stupid question. It was like having a patient teacher on call 24/7.

When ChatGPT Plus launched in February 2023, it was the best £20 I've ever spent, no more getting locked out during peak hours. I remember refreshing Reddit obsessively for GPT-4 rumours, and when it landed the jump felt huge: proper internet access for research, and noticeably better code.

Even into 2024 I was still copying code out of ChatGPT and pasting it into VS Code by hand. I didn't use Git or any real source control, new versions of a project were just folders named "project v1", "project v2" and so on, uploaded with FileZilla. Then I found Cursor, which meant I could ask for a change inside the editor itself instead of copying and pasting.

Later that year I saw a tweet claiming Claude 3 beat GPT-4. I'd never heard of Claude, but I was curious enough to sign up. I was working on a side project scraping cargo tracking data from airline sites into a structured API, built in Python, a language I'd never used. Adding a new airline meant pasting the project into Claude with a sample payload and asking it to implement support. A few issues aside, I was impressed with how well it handled it. Claude became my daily driver from then on: plan with GPT-4, build with Claude.

Where I am now

These days I use Claude Code for most of my coding, with Codex picking up the rest. AI has genuinely been the best teacher I've had. Before it, I'll happily admit I didn't understand deployment properly (I thought everyone just used FTP), Git branching and committing, Docker, splitting a backend from a frontend, or even the terminal, npm, Homebrew and environment variables were a complete mystery.

AI didn't just help me write blog posts. It helped me actually build things, and I'm still learning something new most days.

These days, if I find an app I like the idea of but not the price, the limits, or the way it works, I'll just use AI to build my own version instead. It's become second nature: see something, decide whether it's worth replacing, then go and build it.

Of everything I've started, I've properly shipped one: an iOS app for tracking food in your pantry. I have no Swift knowledge, had never built an iOS app, and had never opened Xcode before. I didn't suddenly become an iOS developer, I just kept going back and forth with Claude Code until it worked. For someone whose old workflow was duplicating folders and calling them "project v1", getting something live on the App Store felt unreal, and it was the first time I'd actually finished one of my own ideas rather than letting it stay an idea.

Given how far this has come in such a short space of time, it's hard to argue we're anywhere near done. What started as testing blog posts turned into learning to ship software properly: Git, environments, APIs, deployment, and eventually an app in the App Store. We're still early. That's the part I find genuinely exciting.