I've never been particularly interested in staying in one lane.

Professionally, I've spent more than 15 years working around a deceptively difficult problem:

How do you help someone know what to do next when the answer isn't obvious?

That question has taken me through healthcare, clinical education, simulation, product design, platforms, AI, organizational leadership, and teams ranging from tiny experiments to organizations of more than 95 people.

It has also taken me into considerably less professional territory.

I used to think these were unrelated interests.

I don't anymore.

Making

I like making things tangible.

I've always been drawn to the moment when an idea stops being theoretical and has to survive contact with reality.

That can mean shipping a product used by millions of people. It can also mean putting two gears together and discovering the mechanism you confidently imagined does not, in fact, work.

Both are useful.

I paint. I work with clay. I build things from LEGO, motors, sensors and whatever else happens to be on the workbench. Some of it ends up on Instagram. I'm increasingly interested in robotics and physical AI because they force software into the physical world.

A robot cannot hide behind a roadmap. The wheel turns or it doesn't. The sensor sees the object or it doesn't. The arm reaches the thing or misses it completely.

Making physical things introduces a kind of honesty that presentations, documents and meetings sometimes allow us to avoid.

The object eventually gets a vote.

That is one reason I keep building.

Writing

I write because some questions need a different instrument.

I'm also writing a novel called It Held. It's a literary near-future story about a family living with increasingly capable AI agents — and about identity, usefulness, parenthood, intimacy, trust and what remains distinctly ours when machines become extremely good at performing pieces of us.

AI is in the novel. But the novel isn't really about AI. It's about people.

That distinction matters to me. Technology conversations have a tendency to become conversations about technology. Capabilities. Benchmarks. Models. Agents. Automation.

Writing fiction forces a different question: What does this actually do to a person?

A dashboard can tell you what someone did. A story can sometimes help you understand why they could not have done anything else.

I've discovered that product work and fiction are both forms of observation. They simply operate at different resolutions.

Repair

I still believe in repair.

One of my stranger hobbies is restoring 1980s boomboxes. They are terrible objects to own rationally. They're enormous. Mechanically complicated. Full of aging belts, oxidized contacts, delicate plastic and electronic components nobody expected someone to be repairing four decades later.

Which is part of why I like them.

There's satisfaction in understanding why something stopped working and bringing it back instead of throwing it away. A dead cassette mechanism stops being junk once you understand the mechanism.

I think repair is underrated. Objects. Products. Organizations. Ideas.

Sometimes the thing really is finished. But sometimes it doesn't need replacing. It needs somebody willing to understand it.

Some problems are just old belts and patience.

Everything else

And then there are the things that don't fit neatly anywhere.

Which may be my favorite category.

Some people have suggested otherwise.

They are wrong.

Side quest

There's also the stand-up thing.

Years ago I gave myself a side goal: perform stand-up comedy on five continents.

Not because I'm secretly trying to become a professional comedian. I'm not. I just like the idea of periodically becoming a beginner at something that offers nowhere to hide.

Stand in front of strangers. Tell them something you believe is funny. Find out.

That is an unusually efficient feedback loop.

I haven't finished this one yet. That's part of the appeal.

Now

Lately, I keep coming back to trust.

AI has made an old product question much more urgent: What should a system be allowed to do on someone's behalf? Not merely: What can it do?

That difference increasingly defines the work I care about.

I'm interested in AI agents, memory, human authorization, identity, voice systems, privacy, robotics and the boundaries between assistance and autonomy.

That exploration has led to Keel — work focused on what trustworthy personal AI infrastructure might look like when systems can remember, act, speak and increasingly operate across the boundary between software and the real world.

I'm much less interested in AI as spectacle than AI as infrastructure. Useful. Understandable. Bounded. Reversible where possible. Worth trusting.

The technology will continue getting more capable. The harder question is what we choose to build around it.

What I'm trying to build

I'm increasingly drawn to the places where disciplines collide.

I want to build technology that expands human capability without quietly removing human agency. I want to keep making physical things badly enough that eventually I make them well.

I want to travel farther. Ride more trails. Write more books. Make more art. Build robots. Eat things I can't pronounce. Keep repairing machines nobody needs. Do stand-up somewhere very far from home.

And keep finding difficult, interesting problems surrounded by people who care enough to solve them properly.

That seems like a good direction.