About

Rheaply changed the order I work in. I spent two years on an asset platform for workplace teams, including a Google engagement that covered 238 buildings full of things nobody could count. I ran 46 research sessions, the most I have done on one project.

Before it, research was something I did because it was part of the process. After it, research was the process. I stopped trying to prove my ideas were right and started looking for evidence that they were wrong. When the first data model failed in implementation, the research gave us enough understanding to replace it instead of defending it.

I also started writing production code because the team was small and helping was more useful than waiting. Nothing builds empathy for engineers faster than spending two hours implementing your own simple design change.

I started weekly company-wide design office hours too. Most of the people in the room were not designers. That is where I discovered how much I like making other people better at the work.

I joined Runwise as the company’s first designer and assumed the job would mostly be designing software. It turned out I was also designing how the company designs. The product controls buildings, and the people using it range from property managers at desks to technicians installing hardware in hundred-year-old basements with no signal.

I work in the open. Every version goes into a company channel, numbered and annotated with the questions I have not answered yet. Engineers, product managers, sales, support, and installers can all argue with it while changing direction is still inexpensive. It feels a lot like architecture school, only the wall is Slack.

Early on, I designed installation flows before watching a technician use them in a real basement. They worked, but later field visits showed me what the screens could not: the physical sequence, connectivity, and the moments when someone had to stop and ask for help. Field validation has been part of the work ever since.

One of the projects I am proudest of never shipped. I spent about a year exploring multi-zone controls and stopped believing in the direction somewhere in the middle. I pulled the usage data myself: 498 people had touched the feature across 334,512 sessions. Then I made the case to stop, to the same product leader who had asked for the designs. Five years earlier I would have defended the work.

The newest of these systems is AI, and I mean designing it rather than merely using it. I design and co-build agents that investigate buildings, read our codebases, and prepare fixes. Nothing they write moves forward until a person opens a test link, tries the working version, and approves it.

That approval gate is the part I would defend hardest. What keeps me interested is not how much a model can do on its own. It is how people understand what it did, decide whether to trust it, and undo it when it was wrong.

My career does not feel nearly as planned as it probably looks written down. It mostly feels like saying yes to interesting problems, being wrong more often than I would like, and getting lucky enough to work with people willing to tell me when I am.

Outside of work almost all of it goes to my daughter. She plays baseball and soccer right now, and I am at all of it. A sideline is the one place I know of where the right move is to keep my mouth shut and let somebody else figure it out. I am not naturally good at that.

The rest is the Chicago Bears, the Chicago White Sox, and the Georgia Bulldogs. Two of those are an exercise in patience. Sports are the only thing I follow this closely that I have no ability to improve, which is probably why I need them.