What a Retired Accountant's Spreadsheets Taught Me About AI
The Setup
My father-in-law is a retired accountant who has spent years hand-transcribing horse racing records into Excel spreadsheets. It’s his hobby. Race names, tracks, jockeys, odds, payoffs, conditions. One file alone covers 123 tracks from 1950. He did it for the love of the sport. He’s not an AI person. Heck, he is not a change-anything person, he held on to his flip phone until Verizon said they won’t support the phone anymore. He was forced into an upgrade. He’s a horse racing person. He knows the trainers, the bloodlines, the races that defined eras. Think of that friend who can walk you through every defining moment of the NFL sport… the key games, the key plays, the players who made it happen, the stats. That’s my father-in-law with thoroughbred racing. Decades of narratives built from pure love of the sport.
My Approach + AI Role
I fed one of his spreadsheets into AI. No agenda, no pitch. Just curiosity about what would happen when AI met a dataset built by someone who genuinely loves the subject. I asked the AI to build a full interactive dashboard from the data that would make him go, “I never thought of it that way”. It came back with nine analytical sections: Jockey rankings, track condition analysis, purse economics, geographic patterns, heritage timelines. Styled to look like a vintage Daily Racing Form chart book. Then my wife and I FaceTimed him so we could watch him open it.
What Actually Happened
He (first struggled to get to the right mail app/inbox) downloaded the file, opened it, and immediately latched onto the narratives. Not the charts or the data visualizations, the way the headlines were written. The way statistics were stacked against each other. The way certain syntheses were framed. He asked me if I’d written it. Not once. Three or four times across this big dashboard. “Did you do this?” I think, looking back on it, he thought I had finally entered his world. That I’d somehow absorbed enough of his passion for the sport to understand the trainers, the rivalries, the history well enough to frame it the way a racing person would frame it. The language was his language. The emphasis was on the things he would emphasize. He started talking about the history behind certain trainers and horses and races the dashboard had surfaced. Connections he recognized instantly. My wife and I had to laugh and say, more than once, “No… this was all AI. We didn’t do much. We just knew what to feed it.” I’ll be honest: most of the content was completely meaningless to me. I don’t know racing. But watching his reaction, it was clear the AI had gotten close enough to something real that it triggered recognition in someone who does know. Whether every detail was perfectly accurate, I can’t say. But the experience of watching someone who wants nothing to do with AI suddenly come alive because the output spoke his language. That was something. His next instinct: “Boy, people would pay money for this kind of stuff. We should build a website!” (So much more on this point in another post).
The Real Insight
The best AI demonstration I’ve ever given had nothing to do with AI. I didn’t teach him a tool. I didn’t explain a workflow. I didn’t pitch the future of anything. I just took something he’d invested years in and ran it through a system that reflected it back in a way that felt like understanding. The technology disappeared completely. What was left was a person feeling like decades of quiet, careful work suddenly mattered to someone other than himself. And that it lived outside an Excel file.
Try This If…
Someone in your life has a passion project, a side obsession, a body of knowledge they’ve been building for years that no one else really gets. Feed it to AI. Not to replace what they know. Not to prove a point about technology. Just to see what happens when sustained attention meets a system that can surface what’s been hiding inside it. You might not understand the output. They definitely will.
Systems Lens
I’ve written before about the identity layer in AI transformation: the gap between teaching people tools and helping them become new professionals. This was a version of that, except there was no transformation program, no L&D curriculum, no proficiency target. Just a retired accountant, a spreadsheet, and the experience of being understood. This random moment made me learn that sometimes the most powerful thing AI can do isn’t to make someone more productive. It’s to give you a new dimension on something you’ve known for years. And once you feel that first ‘what if’ from an angle you never had… you don’t stop. The depth was always there. You were just living in it one way.