About
More than two decades of leading teams and building prototypes. Now I'm testing how far one person can go when AI agents do the research, the drafts and the first builds, and I do the directing.
AI
I've been working with AI longer than it was fashionable. My degree at Warwick included AI research; at the Financial Ombudsman Service I built RPA automation that reduced manual case loading by 80% - the system read, categorised and tagged new enquiries autonomously, with a human reviewing the final 20%. At Future Foundry I led AI proof of concepts including a routine recommendation engine for cosmetics.
At ITG I'm rolling out agentic coding tools across a team of ~50 developers - story point delivery is up by up to 20%, and code quality is improving across all repos. I'm not interested in AI as a buzzword. I'm on a genuine adventure to use AI agents to 100x personal and team productivity - and I think we're only at the beginning of what that means for how digital products get built.
Right now
I've always been the person who builds the demo. It started with "do something cool" at Ford and carried on through the innovation lab at Worldpay and into the platforms I architect today. The hard part was never the idea. It was finding the hours to research it, shape it and get a first version in front of people.
So I'm running an experiment. Can I 100x my output by delegating whole layers of work to AI agents? I don't mean one chatbot. I mean a small team of agents, mostly Claude Code and Cursor, across my day job, my studio and home.
It works in layers. I catch an idea, often as a voice note. An agent researches it and comes back with options and a next step. Another shapes it into a plan, then builds a prototype. I review, redirect and decide what ships. I'm the editor and the director, not the typist.
At work I'm leading an AI enablement (embedding agentic coding tools and AI into development teams to generate real productivity and quality gains) programme that rolls these agentic tools out to our front-end and back-end developers, and we're nearly through it. Some of it works brilliantly: agents are fast, tireless and good at the dull 80%. Some of it doesn't. They need clear briefs, guardrails and someone to say "no, not that". Vague asks get vague work. That sounds a lot like leading people, and it is.
I'll be writing up what I learn as I go, on my personal blog and on LinkedIn.
Making complex ideas tangible, from AR car configurators and LEGO payment demos to generative-AI proofs of concept. Now I prototype at the speed of a conversation.
See the portfolio →Platforms that hold up at scale, from an enterprise marketing platform to a Ford system processing over $3bn a year. The same thinking now goes into how agents, tools and data fit together.
See the portfolio →I founded and ran Digital Technology Labs for seven years and led teams at Ford, Worldpay and beyond. Directing agents draws on the same skills: clear intent, good feedback, knowing when to step in.
Find out more →Claude Code and Cursor are my daily tools, at work and at home: agents research, draft and build, with me as editor. I'm also leading their roll-out to the developers I work with.
See what I'm building →Built with agents
Leading the programme that brings Claude Code and Cursor to our front-end and back-end developers: the tools, the ways of working and the guardrails.
A studio board where every idea is a folder. Agents explore it, shape it and prototype it, then hand it back to me for a decision. They never publish, spend or contact anyone. Only I move things to "launched".
A spare Mac mini drives a screen at home. It recognises family phones on the Wi-Fi and says "Welcome home", with the weather, birthdays and expected guests. Nothing leaves the house.
A shared family wall for the kitchen iPad, with the calendar, reminders and handwritten Apple Pencil notes. One codebase on iPhone, iPad and Mac, built by agents.
A big screen showing which agents are working, what's waiting on me, what's shipped and how many tokens we've used today. Mission control for a team that never sleeps.
Worldpay Innovation Lab
From my time in Worldpay's innovation lab: five granted US patents, assigned to Worldpay, on augmented reality and moving purchases between connected devices. Status: All patents actively held by Worldpay Ltd, expiring 2040.
Granted 24 Feb 2026
Securely moving digital items and transaction details between connected devices, so a purchase started on one can be finished on another.
View the patent (opens in a new tab)Granted 20 Jan 2026
Taking objects out of an augmented reality scene and showing them on another screen, such as a TV, where people can interact with them.
View the patent (opens in a new tab)Granted 24 Jun 2025
Picking up a virtual object in one augmented reality space and putting it down in another.
View the patent (opens in a new tab)Granted 13 May 2025
Moving something you're looking at on one device, like a product or a basket, across to another connected device so the purchase can be finished there.
View the patent (opens in a new tab)Granted 19 Nov 2024
Displaying and interacting with objects from an augmented reality scene on another device, such as a TV or a phone.
View the patent (opens in a new tab)Documents
The full CV, and a downloadable visual portfolio, if you'd rather read the long version.
If you're building teams that work alongside AI agents, or you'd like to, I'd love to talk. I'm open to conversations about leading that kind of work.
Get in touchHiring, a project, or a quick question for my AI.
Chat needs a newer browser: email me instead.
