[[["容易理解","easyToUnderstand","thumb-up"],["確實解決了我的問題","solvedMyProblem","thumb-up"],["其他","otherUp","thumb-up"]],[["缺少我需要的資訊","missingTheInformationINeed","thumb-down"],["過於複雜/步驟過多","tooComplicatedTooManySteps","thumb-down"],["過時","outOfDate","thumb-down"],["翻譯問題","translationIssue","thumb-down"],["示例/程式碼問題","samplesCodeIssue","thumb-down"],["其他","otherDown","thumb-down"]],[],[],[],null,["[See all winners](/competition#w-9) \nBest Game app \n\nPen Apple \nRoguelike deck builder where fun, chaos, and stories are created from player interactions \nWhat it does\n\nPen Apple is a roguelike deckbuilder where AI crafts the fun, chaos, and stories. The game uses Gemini to dynamically generate and interpret in-game content, especially the cards. Instead of pre-programmed logic, natural language descriptions serve as the \"ground truth\" for Gemini Flash to determine card effects. This enables zero-effort creation of complex actions like \"Anagrammer,\" which transforms all cards into their anagrams \n. \nWe've moved beyond traditional procedural generation, using LLMs to guide content creation. Gemini Flash designs the lore, enemies, loot, and rewards, capturing the spontaneous gameplay and accidental absurdity of classic roguelikes and early interactive fiction. \n\n\"Generation\" has always been central to roguelikes and video games, rooted in tabletop RPGs where a game master crafted stories. As RPGs transitioned to video games, procedural generation emerged to replicate this improvisation. While games like Dwarf Fortress have taken procedural generation to new heights, LLMs offer a more natural way to create emergent gameplay. In Pen Apple, we fully embrace Gemini as the primary content generator, leveraging its long context window for deeper storytelling. \n\nBy using a retro game engine, we evoke the charm of the earliest roguelikes while showcasing the cutting-edge capabilities of Gemini Flash. Ultimately, Pen Apple aims to capture the fundamental fun of games through wordplay, chaos, and emergent narratives. \nBuilt with\n\n- Imagen 2 \nTeam \nBy\n\nAtelier Ume \nFrom\n\nUnited States \nMore winners \n[Gaze Link\nBest Android app](/competition/projects/gaze-link) [Outdraw.AI\nMost Creative app](/competition/projects/outdrawai) \n[](/competition)"]]