Best Side

I was standing at a wildly large grill on a beautiful day last Saturday – a balmy 72 Florida degrees, husking my 60th ear of corn, sweating through my shirt, going at it alone, having never cooked for more than 10 people in my life. I was on a team – but they had other duties.

Five teams. Each with a professional chef. Next to me was the team represented by a Food Network legend. Our team? Not a day spent in a commercial kitchen – ever.

The Chi Chi Miguel Throwdown is a 20-year-old charity BBQ competition on the Emerald Coast that has raised over $20 million for children’s charities. Ribs. Side. Flare. Each category gets a winner. The theme this year was “Welcome to the Jungle.”

I was competing for Best Side.

I am not a chef. I have never claimed to be one.

But I had something the other teams didn’t: a recipe that was built, tested, iterated on, and scaled by AI - from a single test batch to 150 competition servings - with me as the feedback engine. It had to be competition grade, AND achievable by a pedestrian cook.

Smoked Habanero Polynesian Creamed Corn.

Pecan-smoked habanero. Cherry-smoked fresh corn. Coconut milk. Heavy cream. Fish sauce. Lime. Tajín. Toasted coconut.

I did a practice run. Gave feedback. Got into sessions to dial it in. The practice run wasn’t hard. What I didn’t plan for was the day-of effort: 60 ears means 60 husks, 180 turns on the grill, 240 cuts for the kernels, 240 scrapes by hand for their milk. That doesn’t even cover the onions and the rest. Then 150+ individual servings: each one plated on request with a spoon, a dash of Tajín, and a crown of toasted coconut… as people came to the tent one by one. Hot. Fresh. Nobody waiting long.

Two judges told me they’d bathe in it.

We won Best Side.

This series has said from the beginning: AI gives you the 80%. Your 20% closes the gap.

This time, I flipped that script. The expertise I was missing wasn’t taste, it was architecture. The coconut milk coats the mouth first, building just enough of a blanket to let the habanero’s heat arrive late and disappear before it becomes intolerable. The cherry and pecan smoke played together like a symphony, as if they were born for corn - and knew they’d be sitting right next to our ribs in the flavor profile. And then the sleepers: Tajín on top - fifty people asked what that spice blend was, convinced there was lemon in it (there ain’t!) - and toasted coconut for a crunch that played against every pop of corn in your mouth.

AI gave the 20%: the recipe architecture, the scaling math, the execution timeline. I brought the 80%… the taste memory, the feedback that shaped every iteration, and the determination to execute something I had no business attempting.

The thesis works in both directions.

The recipe is attached. Go win something.