Treat roughness as evidence, not failure.
The question is about access and feedback speed. Visual polish matters later; the first layer preserves what the rough prototype reveals.
Back to works
02.03.02 — Chromopong
Playable artifact2024 · AI-assisted game prototype
Chromopong
When creation tools become accessible, beginners do not start with productivity. They start with play.
Role
Product observation, artifact critique, and framing around a beginner creation loop.
Audience
Young or nontechnical creators with ideas before technical vocabulary, plus the tools trying to translate that intent into working software.
CHROMOPONGAI BUILD LOOP
ideapromptplay
Primary artifact
Playable AI-built game
01 — Context
What this is really about.
Chromopong is intentionally small. Its value is not polish; it is the creation loop. A 12-year-old with an AI builder can move from idea to playable interaction quickly enough to react, revise, and learn what the tool understands.
Product Question
What does AI building make possible for a beginner, and where does the tool still need to teach product structure?
02 — Problem
The user pressure.
AI can generate a playable surface quickly, but beginners still need structure: rules, feedback, difficulty, progression, and a way to understand why an interaction feels unfinished.
03 — Evidence
What the artifacts prove.
Behavior
The first instinct was play.
Given an AI builder, the user did not reach for a work app. They tried to make a game, which says a lot about what accessible creation tools invite first.
Loop
Playable beats explainable.
A playable prototype gives immediate feedback: speed, difficulty, responsiveness, and whether the idea is fun enough to keep changing.
Artifact
The playable version makes the loop concrete.
The live Replit artifact gives the case study something playable to evaluate: rules, speed, responsiveness, and the next iteration it invites.
04 — Product Judgment
Product decisions.
Frame the user as creator, not tester.
A young user is not only testing a game. They are learning what the builder understands, what it misses, and how to ask for changes.
Show the prompt-to-play loop.
The strongest evidence pairs prompts or notes with screenshots of the resulting gameplay changes.
05 — Evidence Plan
What is still missing.
Capture one complete loop around the playable artifact: initial idea, prompt, generated game, reaction, and one revision.
Gameplay capturePrompt notesIteration loopBeginner friction
Open Questions
01
Which parts of the game idea did the tool understand without extra explanation?
02
Where did the user need adult or product guidance?
03
What would make the next iteration feel more like authorship and less like prompt guessing?