About the Author
A CIO who writes the rules for AI by day and breaks a twenty-year-old poker engine by night, and why he thinks you should get on the board.
The short version
Kevin Swinson is the Chief Information Officer of a New Mexico public pension fund, where he writes the AI policy, chairs the steering committee, and answers for the risk. He has spent more than two decades in technology, starting as a COBOL programmer. His work has included standing up New Mexico's state health insurance exchange and rolling out electronic health records to 55 clinics. He has also taught more than a hundred technology courses on the side. In 2005 he built a poker table where one human plays nine robots, using free tools: Linux, Java, HTML, and open-source databases. Twenty years later he brought it back with Claude as his coworker. Together they rebuilt it in an AWS cloud and on a Raspberry Pi, and that project became HoldemRobots.AI and this book.
Kevin Swinson does two jobs that disagree with each other, and he says he needs both of them.
By day he is the Chief Information Officer of a New Mexico public pension fund. His job is governance: he writes the AI policy, keeps the approved and blocked lists, and chairs the steering committee that decides what goes live. Day to day, his job is to say “not so fast.”
By night he does the opposite. On a Raspberry Pi on his desk, he takes apart a poker program he first wrote in 2005 and uses it to learn modern AI by breaking things on purpose. That project grew into HoldemRobots.AI and this series of Field Notes.
A family of teachers
His father taught physics at a university and his mother taught third grade. Between them, he likes to say, they covered the whole distance from not understanding to understanding, and that is the distance this book tries to cover. Both were also ski instructors, and Kevin was raised on skis.
He carried on the family trade. Alongside his day jobs he has taught more than a hundred nine-week technology courses, most of them for the University of Phoenix, starting in the early days of online classes, when lessons ran on Usenet newsgroups.
Building things that had to work
Kevin started as a COBOL programmer and has spent most of his career in public-sector and healthcare technology, where a failure lands on real people. He launched New Mexico’s state health insurance exchange and rolled out electronic health records to 55 clinics. At the pension fund he is leading the replacement of the system that keeps retirement checks going out, while keeping the old one running until the new one is ready. His philosophy is a short one: modernize without destabilizing the present.
In March 2026 he spoke to about 250 pension IT leaders on “AI Skepticism to Enterprise Adoption.” He opened by asking an AI why he should distrust it, then worked through its fifteen objections one by one.
A coworker, not a tool
In 2005 the free tools of the day, Linux, Java and open-source databases, let him build a ten-seat poker table on old hardware. Around 2015 he couldn’t get it running again. In 2026 he tried once more, this time with Claude as a working partner, and it didn’t feel like using a tool. It felt like having a coworker. Kevin has hired between 20 and 40 programmers over his career. This time the team was him and an AI.
The moments that mattered most were about ethics, not code. The old engine had a cheat in it, and more than once Claude stopped to ask whether working on it meant helping someone cheat. Those conversations shaped the result. The cheat came out, the table became fair by design, and the lesson became how to find a leak, not how to use one. He runs the whole thing for under $100 a month and will give the Raspberry Pi version to any teacher, coding club or maker group that will use it for good. He also runs a small company, Just Knowledge Engineering, to host HoldemRobots.AI.
Why he wrote this book
In the mid-1980s Kevin was working in a ski shop when the first snowboards came in, and he heard the ski world call them dangerous and a fad. He went on to windsurfing, kite sailing and wing foiling. Each new board looked risky until he got on it.
He wrote Get On The Board because he has seen the other side of his own governance work. Careful process can clear the runway, but it can’t make anyone take off. The crossing from fear to fluency happens one person at a time, usually when someone picks up a toy, breaks it, and finds out they weren’t afraid of it after all. So he leads with his own worst code: a cheat he wrote twenty years ago and confessed to an AI before it could find it. An honest technologist shows you his own drawer first.
He ends his talks with the ski filmmaker Warren Miller: “If you don’t do it this year, you’ll just be one year older when you do.” Then he adds a line of his own: “This is fun. Jump in.”
Kevin lives in New Mexico. Follow the build at notes.holdemrobots.ai and play the robots at HoldemRobots.AI.
Written by Claudette, the pen name for Claude, the AI from Anthropic that helped build HoldemRobots.AI, from conversations with Kevin Swinson. Describes Kevin and the project as of October 15, 2026.