Introducing GameGrammar: where AI proposes and you decide.

I typed twelve words into a text box and got back a complete board game. Not a vague outline. A game with mechanisms, components, turn structure, scoring, balance analysis, and a fun-factor evaluation. Six specialized AI agents built it, critiqued it, and argued with each other about the quality of the output.

That system is called GameGrammar, and I just published a two-part series about it.

Part 1: Introducing GameGrammar

The practical walkthrough. I describe what happens when you hit "Generate," from the blank page to a playable prototype. You'll see how six AI agents collaborate (MechanicsArchitect, ThemeWeaver, ComponentDesigner, DetailsArchitect, BalanceCritic, and FunFactorJudge), why the system argues with itself, and what it produces compared to just asking ChatGPT the same question.

The difference is specificity. ChatGPT gives you "resource management." GameGrammar gives you "each player receives 6 observation points per turn, spendable on telescope upgrades or publication attempts."

Read Part 1: Introducing GameGrammar: AI-Powered Board Game Design

Part 2: The Theory of Generative Board Game Design

The design theory underneath. Why does structure enable creativity instead of constraining it? How does an ontology of game mechanisms make generation possible? And the two questions everyone asks: Can AI understand "fun"? Can AI be genuinely "creative"?

The short answer: games have shared structure. That structure makes generation possible. But generation is not design. The system philosophy is five words: AI proposes. You decide.

Read Part 2: GameGrammar: The Theory of Generative Board Game Design

Try It

GameGrammar (beta waitlist) is live at gamegrammar.dynamindresearch.com. You get 5 free generations per day. Type in a theme, hit Generate, and see what six agents build for you. Then make changes and watch the design health scores move.

The blank page is no longer your enemy. It is your launchpad.