stephane.bio
  • Invest
  • Build
  • Write
  • Think
Ketchup
Vibe Building a 2D game with AI
🏏

Vibe Building a 2D game with AI

/tech-category
GamingFuture of work
/type
Content
/read-time

15 min

/test

Vibe Building a 2D game with AI

I’m building a game called Infinity Grid, and I’m doing it with AI at my side. Instead of sitting down with a single engine or toolset, I’m documenting the process of experimenting across different “vibe coding” platforms—those strange, half-magical spaces where you describe what you want, and code, art, or mechanics take shape.

The idea is simple on the surface: a 2D endless platformer full of geometric walls, shifting illusions, and coins to collect. But the real story is what happens behind the curtain—me testing how far AI can stretch in game creation, where it shines, and where it falls apart. Infinity Grid isn’t just a game; it’s an experiment in building with machines, documenting both the breakthroughs and the brick walls.

About

image

Introducing Infinity Grid: The Endless 2D Illusion Platformer

A new 2D platformer is here to melt minds and test reflexes. Infinity Grid drops players into an infinite world of geometric walls, shifting illusions, and deceptive paths. Every level is procedurally generated, meaning no two runs are the same.

Players choose their character, leap through traps, and collect coins scattered across maps that bend the rules of reality. Invisible walls, collapsing floors, and portals keep each moment unpredictable, while minimalist art and smooth mechanics make the challenge feel sharp and addictive.

With unlimited levels, scaling difficulty, and illusion-driven surprises, Infinity Grid isn’t just another coin-grab—it’s a psychological gauntlet disguised as a simple game.

Name Alternatives

  • Infinity Grid
  • Maze of Mirrors
  • Coinveil
  • Shapescape
  • Phantom Walls
  • HoloJump
  • Echo Chambers

Prompt

Create a 2D endless platformer game with the following features:

  • The game world is made of vertical and horizontal walls composed of randomized geometric shapes (squares, triangles, hexagons, etc.).
  • The level generator creates unlimited levels by procedurally generating new wall patterns and obstacles as the player progresses.
  • The player can choose a character (at least 3 different skins/avatars).
  • Player controls: walk left/right, jump, double jump, crouch.
  • Coins are scattered throughout the map; collecting them increases score.
  • Add “illusion mechanics”:
    • Invisible walls that appear when touched.
    • Platforms that disappear when stepped on.
    • Moving walls that shift geometry to confuse the player.
    • Mirrors or portals that teleport the character unpredictably.
  • The art style is minimalist (flat geometric shapes), but movement and collision must feel fluid.
  • Difficulty scales over time: more illusions, faster moving platforms, and denser walls.
  • Basic UI: display score (coins collected), level reached, and character health/lives.
  • Use clean modular code structure (separate player, level generation, and illusion systems).
  • Prefer a popular 2D framework (e.g., Pygame, Godot, Unity2D—choose one and stick to it).

The Challenge

Test all vibe coding platforms with just one prompt for a game, without using any follow-up prompts.

Best 5

  1. Spawn
  2. Dualite
  3. Storm dev
  4. Leap
  5. Capacity

Lovable

image
  • Nice UXUI
  • Start adventure button does not work - the game canvas was not created
  • Lightning fast
  • A lot of errors

Orchids

image
  • very slow
  • it’s not a game, it’s a trippy 2D animation - no controls
  • not functioning

Spawn

image
  • Exceptional user experience and interface
  • Generate 4 games prototypes in 1 go
  • Take a lot of time to generate (approx. 10-20min)
  • 3 games working on 4
  • Easy iterations

Base44

image
  • Nice UI generation
  • CTAs were not working at the first prompt
  • Game map is basic but controls just work
  • Buggy character choose

Cloudflare

image
  • Flawless UI experience
  • Thoughtful Agentic mode ON
  • Generates a blueprint.md file at the beginning of the project creation
  • Fixes automatically errors
  • Took time to go through all the steps
  • Preview was not available

Replit

image
  • Planning before building
  • takes approx. 20min
  • I am not able to confirm my character or start to play
  • Able to continue developing agent mode ON

Rosebud AI

image
  • You can only choose from few worlds
  • Gamy interface
  • Possibility to import 3d prebuilt assets into your game
  • Sound is active

Dualite

image
  • First generation is fast !
  • Game actually works!
  • The characters are not shapes, they differentiate through colours

Storm dev

image
  • Generation very fast
  • Game works properly
  • Flawless controls

Bolt

image
  • Fast generation
  • Characters choosing was not respected, colour versus shapes
  • Game works but weird screen scoping
  • Space controls does not work so Player is stuck

V0

image
  • Preview did not generate → component did not generate
  • Character choosing did not go through so I was not able to play the generated game
  • Too simple UI but very clean

Emergent

image
  • Took over than 30min to generate app
  • Game did not work
  • I had to wake up my agent 3 times to get a working output

Playmix (made with Lovable)

image
  • very simple interface
  • Agent mode ON
  • Error at first go and continuous error

MGX

image
  • Live code generation
  • Broken UI
  • Game not working

HeyBoss

image
  • Agents working and chatting together
  • Landing page? I asked for a game!
  • Broken UI

Mocha

image
  • Nice and simple UI
  • Game controls works
  • Game characters not visible in the game while playing

Macaly

image
  • Very fast output
  • The game is working !
  • Character is not able to jump pass 2 clicks so player is loosing in less than 200m

Anima

image
  • Very fast
  • Clean UI
  • Game not working, meaning it’s just frontend

Deamoy

image
  • UX lack clarity
  • Generated the game and the component live
  • Very cool first version but too basic !

Ludo

image
  • Great to brainstorm a game idea
  • No game what so ever as an output

Gambo

image
  • I had to pay to generate my first game versus advertised in the output.
  • When I actually ended up generating was pretty pleasing.
  • Very good UXUI
  • Simple interface
  • Game working from beginning to end

GDevelop

image
  • Impossible to create the game in the first place
  • The UX is so complex, it took me 20min to figure it out - I was determined.

Lindy

image
  • Fast generation
  • Basic game generation and map
  • Controls working but buggy
  • URL generated to play the game independently

Cosmic

image
  • Preset questions to fine tune the first version of the game
  • Game map buggy where player is stuck at places
  • Flawless collection of coins

Leap

image
  • Game worked with first prompt
  • Very nice UI
  • Multiple levels game with multiple characters
  • Controls were working

Create Anything

image
  • Game design very simple UI and multiple steps to play game
  • Bug at first prompt fixed automatically
  • Buggy game not working properly

Capacity

image
  • Simple UI
  • Buggy jump controls
  • The only output with real floor set in the map

Solid

image
  • Simple UI and color choice
  • Create a profile which is really nice
  • Buggy map
  • Very nice UI map

Blocks

image
  • Game loading forever
  • Error fix did not work

Onlook

image
  • Game preview is empty

Reflex

image
  • Reflex is not able to build this game for me = honest answer at least
  • A simple game version was created but was poorly done

HeroUI

image
  • Error at first prompt that continuously kept going
  • Simple UI
  • Very complex 2D map which is pleasing
  • Bug while playing with no vertical controls
  • Game is not working!

Same

image
  • Error occurred that was fixed automatically
  • I was never able to play the game

Blink

image
  • Clear game play instructions in home screen
  • No shapes for the characters but emojis which could mean the AI was not able to implement
  • AI was able to fix the error but errors kept appearing
  • I had to wake up the agent multiple times
  • No working game

Rocket

image
  • AI made sure to build a web game with the proper framework set as default
  • AI made sure to build the proper screens before writing code
  • 5-8 min of built time
  • Error found that was fixed immediately
  • 2D map buggy, was not able to play any game

ChatGPT

image
  • Code was written in few seconds
  • No preview available unless I install it locally

X AI

image
  • Code was written in few seconds
  • No preview available unless I install it locally
  • Clear instructions for the play time !

Deepsite

image
  • Preview while building
  • Took a while for a working prototype to get there.

Tempo Labs

image
  • Errors in generating player, game canvas and main routing menu

Subframe

image
  • Empty canvas
  • Basic framework and design system presetted

Reweb

image
  • Empty canvas

Scene

image
  • Website not related to my game

MagicPath

image
  • Very nice UI generated
  • Game controls buggy

Trickle

image
  • Very nice and simple to use UI canvas
  • Game controls not working but fixed in 1 prompt
  • No floor in the game canvas

Google Stitch

image
  • Can’t generate 2d game but just UI
  • Mobile game design not a web app user interface

Databutton

image
  • Very simple UI
  • Control works perfectly
  • Buggy after 2 minutes of play time
/pitch

Create a psychological 2D endless platformer using AI for unique gameplay.

/tldr

- The document outlines the development of a 2D game called Infinity Grid, which employs AI to create an endless platformer filled with geometric obstacles and illusions. - Various vibe coding platforms were tested to generate the game, with mixed results regarding functionality and user experience. - The game features procedurally generated levels, character selection, and unique mechanics like invisible walls and disappearing platforms, all presented in a minimalist art style.

Persona

1. Indie Game Developers 2. Educational Institutions (for teaching game design) 3. Hobbyist Programmers interested in game development

Evaluating Idea

📛 Title The "AI-powered 2D platformer" game development project 🏷️ Tags 👥 Team: Game Developers 🎓 Domain Expertise Required: Game Design, AI Integration 📏 Scale: Global 📊 Venture Scale: High 🌍 Market: Gaming 🌐 Global Potential: Yes ⏱ Timing: Immediate 🧾 Regulatory Tailwind: Low 📈 Emerging Trend: AI in Game Development ✨ Highlights: Unique gameplay mechanics 🕒 Perfect Timing: Growing interest in AI-driven gaming 🌍 Massive Market: Gaming industry worth over $200 billion ⚡ Unfair Advantage: Innovative use of AI 🚀 Potential: High engagement and retention ✅ Proven Market: Established gaming audience ⚙️ Emerging Technology: AI, Procedural Generation ⚔️ Competition: Medium 🧱 High Barriers: Custom technology development 💰 Monetization: In-game purchases, ads 💸 Multiple Revenue Streams: Subscriptions, one-time purchases 💎 High LTV Potential: Engaged user base 📉 Risk Profile: Moderate 🧯 Low Regulatory Risk: Minimal legal implications 📦 Business Model: Freemium 🔁 Recurring Revenue: Yes 💎 High Margins: Digital sales 🚀 Intro Paragraph Infinity Grid is an AI-powered 2D endless platformer that leverages procedural generation to create unique gameplay experiences. With monetization strategies like in-game purchases and subscriptions, it taps into the rapidly growing trend of AI in gaming. 🔍 Search Trend Section Keyword: "AI in gaming" Volume: 60.5K Growth: +3331 📊 Opportunity Scores Opportunity: 9/10 Problem: 8/10 Feasibility: 7/10 Why Now: 9/10 💵 Business Fit (Scorecard) Category Answer 💰 Revenue Potential $1M–$10M ARR 🔧 Execution Difficulty 6/10 – Moderate complexity 🚀 Go-To-Market 8/10 – Organic growth through gaming communities ⏱ Why Now? The surge in AI technology and gaming popularity makes now the perfect time to innovate within the gaming sector, attracting both avid gamers and new users. ✅ Proof & Signals Keyword trends show significant interest in AI applications in gaming, supported by Reddit discussions and high engagement on Twitch and YouTube channels featuring AI-driven games. 🧩 The Market Gap Current gaming experiences often lack personalization and dynamic challenges. Infinity Grid addresses this by providing procedurally generated environments that keep players engaged and constantly surprised. 🎯 Target Persona Demographics: Ages 15-35, gamers, tech enthusiasts Habits: Regular gaming sessions, active on social media Pain: Desire for novel gaming experiences without repetitive gameplay 💡 Solution The Idea: Infinity Grid is a 2D endless platformer that uses AI to generate unique game levels and challenges dynamically. How It Works: Players navigate through procedurally generated levels filled with geometric obstacles and illusions, enhancing replayability. Go-To-Market Strategy: Leverage gaming forums, social media ads, and influencer partnerships to build awareness and launch the game. Business Model: Subscription model with in-game purchases for cosmetic upgrades and special features. Startup Costs: Medium Break down: Product development, marketing, and legal setup. 🆚 Competition & Differentiation Competitors: Other indie platformers and AI-driven games Intensity: Medium Differentiators: Unique procedural generation, minimalist aesthetic, and engaging illusion mechanics ⚠️ Execution & Risk Time to market: Medium Risk areas: Technical challenges in AI implementation and market competition 💰 Monetization Potential Rate: High Why: Strong player retention through engaging mechanics and monetization strategies. 🧠 Founder Fit The founder's background in game design and AI technology positions them well to lead this project. 🧭 Exit Strategy & Growth Vision Likely exits include acquisition by a larger gaming studio or an IPO. The vision includes expanding to mobile platforms and creating a suite of games with interconnected mechanics. 📈 Execution Plan 1. Launch a beta version to gather user feedback. 2. Utilize SEO and community engagement for acquisition. 3. Optimize conversion through targeted offers. 4. Scale with referral programs and social sharing features. 5. Aim for 1,000 paid users within the first year. 🛍️ Offer Breakdown 🧪 Lead Magnet – Free demo version of the game. 💬 Frontend Offer – Low-ticket introductory skins and features. 📘 Core Offer – Full game access with subscription model. 🧠 Backend Offer – High-ticket consulting for game development insights. 📦 Categorization Field Value Type Game Development Market B2C Target Audience Gamers Main Competitor Indie platformers Trend Summary AI-driven gaming experiences are becoming increasingly popular. 🧑‍🤝‍🧑 Community Signals Platform Detail Score Reddit Active gaming subs with 2.5M+ members 9/10 Facebook Gaming groups with 200K+ members 8/10 YouTube Many relevant creators showcasing AI games 8/10 🔎 Top Keywords Type Keyword Volume Competition Fastest Growing "AI platformer" 30K MED Highest Volume "Endless runner game" 100K LOW 🧠 Framework Fit The Value Equation Score: Excellent Market Matrix Quadrant: Category King A.C.P. Audience: 9/10 Community: 8/10 Product: 9/10 The Value Ladder Diagram: Bait → Free Demo → Core Game → Premium Features ❓ Quick Answers (FAQ) What problem does this solve? It addresses the need for dynamic and engaging gameplay experiences. How big is the market? The gaming market is worth over $200 billion. What’s the monetization plan? Freemium model with in-game purchases. Who are the competitors? Indie platformers and established gaming franchises. How hard is this to build? Moderate complexity with a focus on AI and procedural generation. 📈 Idea Scorecard (Optional) Factor Score Market Size 9 Trendiness 8 Competitive Intensity 7 Time to Market 7 Monetization Potential 9 Founder Fit 8 Execution Feasibility 7 Differentiation 8 Total (out of 40) 63 🧾 Notes & Final Thoughts This project leverages an emerging trend in AI and gaming, providing a unique opportunity to capture a growing market. The potential for high engagement and revenue makes it a solid investment opportunity, but technical execution will be critical.

User Journey

# User Journey Map for Infinity Grid ## 1. Awareness - User Trigger: Discovery through social media or word-of-mouth. - Action: User sees a promotional video or post about Infinity Grid. - UI/UX Touchpoint: Engaging visuals and gameplay clips on social media platforms. - Emotional State: Curious and intrigued, but skeptical about its uniqueness. ## 2. Onboarding - User Trigger: User decides to try the game after seeing positive feedback. - Action: Downloads the game and completes the initial setup. - UI/UX Touchpoint: A simple tutorial introduces game mechanics and controls. - Emotional State: Hopeful yet anxious about complexity. ## 3. First Win - User Trigger: Successfully completes the first level. - Action: User receives immediate feedback and rewards (coins collected). - UI/UX Touchpoint: Celebratory animations and a score display. - Emotional State: Excited and accomplished, fostering a sense of achievement. ## 4. Deep Engagement - User Trigger: User becomes immersed in gameplay, seeking to beat their high score. - Action: Regularly plays, exploring different characters and levels. - UI/UX Touchpoint: Dynamic level generation keeps the game fresh; leaderboard displays. - Emotional State: Engaged and invested, feeling a sense of progression and challenge. ## 5. Retention - User Trigger: User receives notifications about new levels or updates. - Action: Returns to the game after a break due to new content. - UI/UX Touchpoint: Email or in-app notifications highlighting new features. - Emotional State: Rejuvenated and eager to explore new content. ## 6. Advocacy - User Trigger: Positive experiences lead to sharing the game with friends. - Action: User posts about the game on social media or recommends it. - UI/UX Touchpoint: Easy sharing options within the game interface. - Emotional State: Proud and enthusiastic, feeling part of a community. ### Critical Moments - Delight: The first win and dynamic level generation that surprises users. - Drop-off: Confusion during onboarding or frustration with difficult levels. ### Retention Hooks - Habit Loops: Daily challenges that reward consistent play and leaderboard rankings. ### Emotional Arc Summary 1. Curiosity: Initial intrigue about the game. 2. Anxiety: Concerns about complexity during onboarding. 3. Excitement: Joy from achieving the first win. 4. Engagement: Deep involvement in gameplay. 5. Pride: Advocacy and sharing with others.

stephane.bio

Made with Notion, Published on Super - 2026 © Stephane Boghossian

LinkedInInstagramMediumGitHubXBehanceDiscordPinterest