Grain to glass
A basic pale ale, scrolled through a fourteen-machine brewery at a scale I've never brewed at. Really an experiment in guiding an AI drawing skill, and in what it could do for teaching.
What hooked me
I've brewed some very complicated beers as Fist Mash Brewing, but never at commercial scale. I'd been trying a new AI skill that draws technical things as isometric machines, and I wanted to see how far I could push it with a subject I know well.
What I built
A full-page scroll through an isometric brewery. One deliberately basic recipe, a 20 hL pale ale, moves through fourteen machines from the steep tank to a pallet of cans. The camera rests on each machine while it works, then follows the batch down the line.
- The process: malting, milling, mashing, lautering, the boil, whirlpool, chilling, yeast pitching, fermentation, a hydrometer reading, carbonation and canning.
- The numbers: every reading on the way comes from a small model of that step, and a thermometer on the panel follows the batch from the steep to the boil and back down to 2 °C.
- Sound for every machine, synthesized in the browser, and autoplay that brews the whole batch in about two minutes.
- The Fist Mash mark is redrawn as an animated vector logo, and turns up on the wall, the tanks and the cartons.
What I learned
The learning wasn't about brewing; it was about the skill and how to guide it. It drew beautifully but didn't know the process. It piped malt from the kiln to the mill, left a water line starting from nothing and gave the mash tun no hot liquor. Each fix came from me knowing how a brewery actually works and saying so. The tool brings the drawing; the expertise still has to come from someone who knows the subject.
That's what makes me think about it pedagogically. A model like this, where every number is real and every pipe has to land somewhere, is a good way to teach a process: students could check it, break it and correct it, the same way I did.