Mark Rochefort

Mark Rochefort

Co-founder and Product Director at Measurelab. Currently raising a small swarm of AI agents. They're learning. So am I.

Twenty-seven years on the web, with a psychology degree and an MSc in Computer Science.

What I'm up to right now:

The web has only felt this interesting to me once before: the late 1990s. I was studying Computer Science at Bristol, writing about autonomous software agents and wondering what the web would become. That feeling is back.

I've spent 27 years building digital things. The early years included BBC Webwise and interactive television work recognised with a BAFTA. I then led technology at Harvest Digital for eight years, where the most interesting problems turned out to be about marketing data and how people make decisions.

I co-founded Measurelab with Dara Fitzgerald in 2013 and still help lead it today. We built it from nothing into an established consultancy. Along the way, annual revenue passed £2m and the team reached 20 people. We have worked with organisations including Google, Salesforce, Lush and Pret.

Today, as Product Director, I'm turning how we operate and deliver into product - an internal operating platform, and a client environment where the work is visible. Both are aimed at making our expertise repeatable without pretending the judgement can be automated away.

In 2023, we began building Brian, an all-company conversational AI agent. It taught us what makes agents useful and how much care is needed before trusting them with real company data. That work now informs both the products I build and how I use AI myself.

I think the future of expertise is collaborative. Skilled people provide judgement while software and agents handle more of the machinery. The interesting part is designing the join: what to hand an agent, what to keep human, and how a small team gets real leverage out of that rather than a pile of half-trusted output.

Which is why I'm raising a swarm rather than reading about them. Some of it works. Some of it is a mess. That's the fun part.

How I work:

I tend to connect the dots across product, technology, data, people and commercial reality - making the wider system understandable, identifying the decisions that need to be made, and helping turn uncertainty into something practical that can be tested.

That often begins as a conversation. Sometimes it becomes a prototype, a clearer plan or a new capability inside the organisation. The aim isn't to make people dependent on me; it's to help them see the system, move forward and increasingly own it themselves.

Other things:

Things I'm thinking about:

๐Ÿ› ๏ธ tech-enabled managed services โ™ป๏ธ repeatable expertise ๐Ÿค– agent-supported operations ๐Ÿงฑ trusted data foundations ๐Ÿค human judgement ๐Ÿงฉ MCP and memory ๐Ÿ“ˆ marketing AI ๐ŸŒŠ systems that create freedom