Operations first. Technology second.

I've sat on your side of the desk.

Most AI consultants arrive from the technology side: they know the tools, and they learn your business as they go. I came the other way. A decade inside regulated financial services businesses taught me how work actually flows through an operations team — where the re-keying happens, which spreadsheet everyone is quietly afraid of, what the FCA expects you to be able to evidence, and why "just automate it" is never that simple when client money and client data are involved.

That background is why my work starts with the process, not the software. The technology changed in 2023; the discipline of running a defensible operation didn't.

Start at the smallest honest level.

Every problem I look at gets classified before anything gets built. Some problems are education — your team already owns the tool, and needs fifteen minutes and a decent prompt. Some are integration — two systems that should be talking, connected in days on a platform you already pay for. A few genuinely need a custom build — and then I write the code. Selling you the third when you needed the first is how AI projects get their bad reputation.

• Tools you already own come first. New software has to earn its place.
• Your team is trained as the system ships. You run it after I leave.
• Compliance is a design input, not an afterthought. Data governance, audit trails, and encryption are decided at the start.
• Honest scoping. If the answer is "you don't need me for this," that's the answer you get.

I build in the open.

Everything I recommend, I practise. Labs shows the products I've built with AI agents — from a secure client fact-find platform for financial advisers to a fully automated editorial desk. And the Journal is where I write — most days — about AI, automation, and regulation for advice firms and other regulated businesses.