Pour Decisions: How I Built a Drink-Shop Game to Teach the Branches of Government

My football fans love the NFL-themed sims I build. This time I wanted to answer a harder question: what kind of game would students who don't play video games actually want to play? The answer turned out to be a drink shop, and a government lesson came with it. Pour Decisions went from a loose idea to a finished, tested game in one afternoon.

The brainstorm that missed. I started with Claude, the AI assistant I work with, and asked for a fun game for non-gamers. Claude suggested designing around what non-gamers tend to enjoy: story, choices, creativity, and low-pressure play with no fast reflexes. I didn't like the list. The ideas weren't bad, but they felt more like activities than games. One still planted a seed: the business sim.

The spark. My students are obsessed with drink shops. A 7 Brew drive-thru opened in town this summer, and they love energy drinks and flavored drinks. Custom drinks have wild recipes. I already teach that a law is like a recipe, the same analogy behind Cookie Court. That's when a fun game became an academic game.

The mapping. Writing the recipe is the legislative branch, like Congress writing a law. Making the drink is the executive branch, like the President carrying out the law. Sloppy work makes people unhappy, even with a good idea. Customers taste the drink and judge it, like courts judging whether a law holds up. The customers also teach representative government. Listen to what they want and they support you. Ignore them and sales fall, like a representative who loses the next election.

Talk first, then build. My rule for Claude was simple: "don't code yet, let's talk it through." We settled the rules and strategy before any code existed.

Three days in one class period. Each student opens their own drink shop and runs it for three in-game days in 5 to 10 minutes of class. Day 1 is a cold open with no clues. Students build any drink they want and mostly guess wrong. That's what happens when representatives don't listen to the people. Day 2 brings clues about what customers want. Day 3 brings new clues, because public opinion changes over time.

The clues. Three sources hint at three hidden ingredients: overheard chatter in line, a social media trend, and the rival shop's menu board. The first clues were too easy, because some named the ingredient outright. Now Day 2 gives one easy clue and two hard ones, and Day 3 gives one easy, one hard, and one hardest. Which source gets which level is random, so there's no pattern to learn. Some of the hardest clues sneak in history, like the tea dumped at the Boston Tea Party.

The five-ingredient cup. One hidden ingredient brings okay sales, two brings better, and three brings great sales. Extras are allowed, but the cup holds only five ingredients, so nobody can dump in the whole shelf. The shelf has about 30 ingredients in five tabs, so guessing without the clues is nearly impossible. A clash rule cuts sales when a drink mixes things that don't go together, like lemonade with milk. Students learn about clashes only from customer reviews, so they have to read the feedback and adjust.

Built for touchscreen Chromebooks. The touchscreens can't sense pressure like a game controller, so every skill game relies on timing and accuracy. When I tested the first version, hold-to-pour worked, but the pump and drag games were too easy. Ingredient pictures also made it look like flavor mattered, so now there are three pure skill games with no pictures. In Pour, students hold a button and let go when the cup fills to a green zone. In Pump, my design, three to five color-coded bottles labeled with keyboard letters light up one at a time, and students press the matching key fast. In Shake, also mine, students swipe back and forth until "STOP" flashes at a random moment. Stopping too early or too late costs points.

Scores go to the spreadsheet. Students sign in at the start, so nobody forgets to submit. Results export automatically to my class spreadsheet, including total sales, each day's sales and correct ingredients, quality, clashes, and a star rating. The final report compares Day 1 sales to Days 2 and 3, so students see that listening to the people paid off.

Who built what. Claude built the base code and tested it by playing all three days in a test browser, checking scoring and the export. I played it, flagged what felt too easy or confusing, and redesigned pieces. Then Grok Build, an AI coding tool, did the visual code. Claude wrote it a handoff document listing what to restyle and what never to touch, like the game logic and the export. The polished version goes back to Claude for one more full test before students see it.

What I'd tell another teacher. A brainstorm is useful even when it misses. Claude's first ideas didn't stick, but one sparked the real one. Start from what students already care about, and the lesson feels natural instead of forced. Talk before coding, and use each tool for what it does best.

You can play Pour Decisions for free on the games page. Try it yourself before you use it in class, and let me know how your students do.

Want to help your teachers save time with AI? Book a PD session for your school.

New ideas every week: Follow The AI Teacher Lab on Facebook.