Skip to content
Coffee Tokens
MenuClose
Meet Michael

I help people understand AI well enough to use it on their own problems.

My favorite question in the world is “how can I use AI to do this?” It usually comes from someone who has been told AI is either magic or a threat, and who suspects it is neither. They are right. It is a tool, and like any tool it rewards understanding and punishes guessing.

A ceramic coffee mug and a closed notebook on a wooden table

Now

By day I’m Director of IT for a multi-brand consumer products company in Northwest Arkansas. We use AI for real work, and we keep it under control.

The other half is Coffee Tokens. I help small businesses find out how AI describes them, and then fix it.

Where I come from

My wife Beverley is my high school sweetheart, and Northwest Arkansas is home. When we’re not working, we like finding ways to be around the water. There is something about the ocean and the sound of the waves that slows everything down.

I spent twenty years in the Navy. The first eight were as an enlisted sailor in naval nuclear power, before the Navy commissioned me from the ranks; I finished as a Lieutenant Commander. Nuclear power teaches you one thing for life: the procedure exists because someone already found out what happens without it. After the Navy came IT leadership in federal healthcare and defense, from a VA medical center to the Department of Labor.

My father was an Air Force pilot, and I got my love of flying from him. I’m a private pilot, I built an airplane, and I’m building the second one now. A few thousand rivets teach you what a project plan never will.

Michael Gross in his office
Michael and Beverley at a promotion ceremony
With Beverley at my brother’s promotion ceremony, 2026.
A homebuilt airplane under construction in a hangar
Airplane number two, in the hangar.

Same job every time: take something that isn’t working well and fix it without breaking the budget or the results.

How I think

  • Map the process before you touch the system.
  • When something breaks, look where two people, systems, or AI agents hand off. That’s where it almost always fails.
  • Govern AI before you scale it. It earns access to real systems the way people do: step by step, with limits.