Building a human-centred practice inside a freight railway.
I joined CPKC in July 2025 with a brief to democratize design at a twenty-thousand-person freight company that had no designers.
We work alongside change management and business transformation, the teams already responsible for how work gets done. What was missing was anyone asking how it felt to use.
Four kinds of work came out of it. A tool, a practice, a method, and the ground they all stand on.
Most of this work is confidential, but message me or find me on LinkedIn if you want to talk through any of it.
Something people use
Six use cases
We talked to people across every line of business. Six ideas for AI came out of it, ranked. This was the one that got built.
Two directions
One version looked like the spreadsheet people already used. The other let them ask a question. The familiar one would have kept them in the spreadsheet.
Shipped rough
Internal tools here were built on default templates. We gave this one a look of its own, then showed it to people while it was still rough. It is in use now, and they run it themselves.
Something that persists
People already doing it
People in other teams were already talking to users and testing things before spending money. They just had no name for it, and no way to find each other.
A group, not a document
We could have written a framework. Most go unread. So we found those people and gave them somewhere to meet.
105 members
A group that grew to 105+ people across Information Services by early 2026. That is how many joined, not how many changed the way they work.
Something that shows shape
No map existed
Freight passes through more hands than any one person can see. Nobody had written down how the work actually moves.
Mapping what happens
We mapped how work moved through an external trucking partner, and later through a customer portal. Both started from a blank page.
Mapping backwards
We asked an engineering team to picture their job a few years out, working the way they wanted to, then traced the steps back from there. Asking what hurts gets you complaints. Starting from the end gets you a plan.
Something underneath
What's already there
Years of customer research already sit in one place, and a new customer portal is being designed on top of it. I am not leading that work. My part was interviewing the internal teams the portal affects.
Rebuilt on it
One early prototype looked finished, but the customers and locations in it were made up. It demoed well. I rebuilt it from the real research.
What comes next
Three documents set the rules for how research gets filed, and they had stopped matching each other. I wrote up a fix and put it in for review.
