From his dad's jobs to AI Metric

Chris got into construction the way the industry used to make people: as a kid on his dad's small jobs, passing tools, mixing badly, measuring twice because his dad checked. It was a hobby before it was ever a career, and it left the lesson that outlasts every technology on this site: buildings get built by people who trust each other, and the paperwork exists to serve them, not the other way round.

The hobby turned into weekend and holiday work with local contractors, and the interest stuck. It became formal when he joined Shepherd Construction as a student, learning the job properly around his studies. That is where the yellow notebook comes in: an old setting out engineer with a hardback Chartwell, pile positions worked out by hand, trigonometry line by line, every coordinate checked before anyone was allowed to trust it. Nothing went on site because the computer said so. It went on site because someone could show you why it was right.

That is a slow way to work, and that was the point. When you have earned a number the hard way, you develop a feel for the wrong ones: the level that does not smell right, the dimension that contradicts the drawing two revisions ago. The tools have changed beyond recognition since then. That instinct has not, and it is worth more now, not less, because answers arrive faster than ever and the checking discipline has to keep up.

After university the career ran through major Tier 1 contractors around the UK: more than twenty years of delivery, from setting out through running packages to the commercial side. Complex facades, building envelope, valuations, variations, NEC and JCT contracts, and the occasional dispute where the only thing that mattered was what had been written down at the time. Long enough on real projects to know exactly which evenings get eaten by paperwork, because for years those evenings came out of his own diary.

The technology side was never a career move either. It was a second hobby that got out of hand. The honest word is nerd: the sort of person who automates a spreadsheet on a Sunday for the satisfaction of watching Monday take care of itself. For years that stayed private, small scripts and shortcuts built for one person, because the day job was building, not software.

Then the models got good, and the hobby collided with the day job. The realisation was not that AI is clever. It was simpler and more uncomfortable: good people, skilled people, were spending their evenings doing work a machine should be doing. Diaries typed up at nine at night. Photos filed by hand. Meeting notes rewritten from memory. Records assembled months later for a claim, at ten times the cost of capturing them properly on the day.

AI Metric exists because both halves of that story are real. The site half knows what a method statement is for, why the record wins the argument, and which tools a site team will quietly refuse to use. The nerd half knows what the models can genuinely do, what they cannot be trusted with, and how to wire them into the systems a business already runs on. Most firms get offered one half or the other. The gap in the middle is where the wasted evenings live.

The yellow notebook is still the standard. Understand the process before you optimise it. Check the output before you trust it. And measure the result in the only unit that ever mattered: whether the people doing the work got their time back.