Millwright

About Millwright

The trade is old. The tools are new.

For centuries, a millwright kept a mill’s machinery turning: the wheel, the shafts, the gears. Every business runs on a machine too, one nobody ever drew: the way work comes in, gets passed along and goes back out. Millwright comes in, finds the parts that grind, and fits AI to them, with your people at the controls.

mill·wright/ˈmɪl.raɪt/noun

  1. A tradesperson who installs, maintains and repairs the machinery that keeps a mill running.

    From mill + wright, a maker. Since the 1400s.

  2. A firm that does the same for a business: fitting AI into the everyday work that keeps it running, and keeping it running well.

    That's us.

Part for part

ThenA water millThe river does the work. The millwright keeps the machinery true.
NowA business, todayThe work comes in. Millwright fits the machinery, and keeps it true.

The team

Millwright is two AI engineers who have designed, built and shipped AI systems for other people. We still do every part of the job ourselves: the design, the code, the fitting and the upkeep.

  • You talk to the people who build it.

    No sales layer and no hand-off. The two people on your first call are the two who design it, write it and fit it.

  • We’ve done this for others.

    We’ve built and shipped AI work for clients before, so we know what holds up in a real working day, and what doesn’t.

  • Both of us, on all of it.

    We don’t split the job into yours and mine. We both design it, build it and check it, so every part gets the experience of both of us.

HereThe workshopTwo engineers at one bench, both of us on every part of the job.

On the board

In a good workshop each tool hangs on a painted outline, so you can see at a glance what’s there. These are the ones we reach for.

  • System design

    Working out what to build, and what to leave well alone.

  • Language models

    Choosing, prompting and tuning the right model for the job.

  • Retrieval and search

    Getting answers from your own documents and data, with the source to hand.

  • Integrations

    Connecting AI to the email, files and software you already run.

  • Evaluation

    Testing outputs against real examples before anything goes live.

  • Security and privacy

    Keeping sensitive data where it belongs, and out of where it doesn’t.

  • Monitoring

    Watching quality, speed and cost once it’s running.

  • Upkeep

    Updating it as the models, the tools and your work change.

Come and say hello.

The first conversation with Millwright is free, and there's nothing to prepare.