Refit AI
My consulting practice, for people, companies, and institutions working out what AI means for them. Electricity paid off once factories were rebuilt around it, and I think AI will too.
Refit AI is my consulting practice. I work with people, companies, and institutions on what to do about AI, and then I help them do it.
The line shaft
The first factories to get electricity mostly used it to replace the steam engine. One big motor sat where the engine had been and turned the same overhead line shaft, which drove the same belts and the same machines in the same order. The building didn't change, so the output didn't either.
The gains came later, when people built factories around the new power: a small motor on each machine, one story instead of several, and machines placed in the order the work moved. The economist Paul David traced this in "The Dynamo and the Computer." Edison's first power station opened in 1882. Factory productivity jumped in the 1920s. The decades between went to rebuilding the factory, not to waiting for better motors.
AI is the new motor
I think the next ten to twenty years of AI will run the same way. Most AI strategy today is a bigger motor on the old line shaft: a chatbot added to an existing workflow, a copilot inside an existing tool, the same org chart with cheaper parts. That helps a little. It isn't where the gains are.
The gains come from asking what the organization would look like if you designed it today, knowing what the models can do. Which steps exist only because thinking used to be expensive? Which roles were built to carry information that now carries itself? What stays scarce, and who should own it? My talk When thinking gets cheap works through that last question.
A refit is what a shipyard does to a working ship. The hull stays. What's inside gets rebuilt for the power it runs on now.
The work
The practice takes four shapes:
- Build. AI systems built inside the tools a team already uses, and put into daily use.
- Lead. Fractional CTO or CPO work for companies between their first product and their first platform.
- Advise. Paid working sessions on one question, build-or-buy reviews, and AI opportunity assessments.
- Enable. One-on-one AI coaching for executives, and workshops for leadership teams.
Most engagements start small. I learn how the work happens today, whether that's one person's week or a whole team's. You get a short list of what to change, ranked by what it's worth, and you pick where to start. Then I build it, set it up, or coach you through it until it's part of the everyday work. Then the next one.
I've spent my career building products and leading the teams that build them, at Two Sigma and at CoinDesk, where a platform rebuild took the site from 1 million to 28 million monthly readers. Most of what I build now, I build with AI agents. The rest of this site is that work.
If you're working out what AI means for your work, your company, or your institution, write to me at hello@refit-ai.com.