Tried and tested.
Now automated.

Bespoke senior AI consulting for teams who know they need to do something with AI, but aren’t yet sure what. Maybe you’ve run a few pilots that quietly went nowhere, or you’re staring at the budget line and don’t yet know where it should go. The field is still young, every business I’ve worked with has needed something a bit different, and most of the value is in the figuring-out part. So we work that out together, and then we build.

I spent 20 years in search and digital strategy (I founded Builtvisible, later acquired), and for the past few years I’ve been building AI infrastructure full-time. I’ve published 16 open-source MCP servers on npm, I run two paying SaaS products entirely on the architecture I’d build for you, and I use every tool I recommend to clients on my own projects before I suggest it to anyone. If it doesn’t work in practice, I don’t sell it.

MIT’s August 2025 study of enterprise AI found that 95% of GenAI pilots fail to deliver measurable financial impact. Only 5% cross what the researchers call the “GenAI Divide” into real production.

McKinsey’s 2025 State of AI reaches the same conclusion from a different angle. About 65% of organisations use generative AI regularly, but only a third have scaled it. The ones that succeed share one thing: they redesigned the workflow first, then applied the AI. The ones that fail bolt a model onto a process that wasn’t ready for it.

That redesign work is the harder, less glamorous part, and it’s where most engagements either stall or skip the step entirely. A strategy deck, a workshop, or a platform subscription can each be useful in the right place. None of them, on their own, do the specific reverse-engineering that moves the number for your business. That’s the work I like to focus on.

Every engagement starts with me spending real time inside your operation. Not a kickoff call and a discovery survey. I watch how your team really spends their days, the spreadsheets they keep open, the systems they copy data between, the Friday morning report someone builds by hand because the tooling never quite worked, the Slack channels where decisions get made before they hit the meeting.

The procedures that are tried-and-tested at the human level are the ones that automate reliably. If your marketing team has been producing briefs the same way for two years and it works, that’s a procedure I can replicate with AI in the loop, faster, more consistent, and off your team’s plate. If something is still being figured out, no amount of AI will make it work until the humans figure it out first.

Once I understand the shape of the business, I build the systems. Document processing pipelines, research and monitoring workflows, content production systems, custom MCP servers that connect AI to your internal data, reporting automations that replace the Friday morning spreadsheet. I build with tools I use every day.

The goal is that six months after I leave, your team is running AI-enabled procedures that feel like a natural extension of how you’ve always worked, just faster, more consistent, and without the manual grunt that used to eat half a person’s week.

Every business is different under the hood. The way your sales team triages leads, the spreadsheet your ops manager has been quietly maintaining for three years, the report someone rebuilds every Monday because the system never quite did it properly. None of that is in a textbook, and none of it is going to be solved by buying a SaaS subscription and hoping for the best.

So a lot of what I do is research and development, in the proper sense. We try things, we measure what happened, we throw out what didn’t work, and we keep what did. The bits that survive get built into something your team can run without me. That’s the bit nobody else seems to want to do, and it’s the bit that moves the number.

If you’ve ever felt like the AI conversation in your business has been all talk and not much shipped, this is probably why. The talking is easy. The R&D is what gets you to the 5%.

AI in business is still embryonic. Most of the agencies and teams I talk to don’t need a product bolted on. They need someone sitting with them, watching how the work really flows, and teaching the fundamentals so the team can keep going after I’ve left.

Think of it as a forward deployed engineer rather than a consultant with a deck. I spend time inside your business, learn the specific shape of it, and we build what fits. No two engagements have looked the same, and I’d be suspicious of anyone telling you otherwise. Past work has included full content production pipelines, bespoke document processing systems, custom AI access to internal databases via MCP servers, AI-driven competitive monitoring, private on-premises AI for regulated data, and senior AI advisory for leadership teams figuring out the strategy layer.

Engagements usually run a few months rather than a few weeks. Long enough to move something real, short enough that you’re not locked in. If what you need is off-the-shelf or urgent-next-week, I’ll happily point you somewhere better. I’d rather have a friendly conversation and send you elsewhere than take work that isn’t a fit.

Pricing is scoped to the work once we both understand it. Talk to me first. Numbers without context aren’t useful to either of us.