Create a playable game by giving the coding agent a clear player, fun target, and test loop.
Source episode · 05:25
How I AI: My First Impressions of Claude Opus 4.8 – Coding, Strategy, and Where It Shineswith Claire Vo
05:25 to 07:03: Claire asks Opus 4.8 to propose and build a game for a nine-year-old, plays the result, and pushes the first version toward a more ambitious 3D design.
A playable game that the intended player can understand, control, and enjoy.
Anthropic AI assistant
Step by step
Follow the sequence once, then adapt the prompts, checks, and handoffs to your own setup.
4 steps
Start a chat in Claude Opus 4.8. Ask for creative, one-shot project ideas, specifying an audience or purpose to guide the model's suggestions.
Return five game concepts with one mechanic, one win condition, and a reason a nine-year-old would replay each.
Review the suggestions from Claude. Look for an idea that includes an ambitious, agentic workflow, such as a process where the AI builds, tests, and refines the project itself. Select the one you want to pursue.
Use a clean coding workspace with browser preview and permission to run the game.
Instruct Claude to build the game or project it suggested. For the most agentic result, you can copy the model's own proposed workflow and use that as your prompt.
Build the selected game, play it through the visible interface, and tune difficulty until a first-time player can finish one round.
Play the initial version of the game to test its functionality and fun factor. Once you have a working baseline, push the model's capabilities by requesting a significant enhancement, like converting the game from 2D to 3D or adding new mechanics.
Record launch, controls, win and loss states, then fix the highest-impact play-test problem.
Turn an idea into a PRD, user stories, and a plan.
After the steps
How to recover when the loop fails and where human judgment helps.
Recover
Keep building
Use a hybrid AI workflow to bridge design and development. Generate high-level design concepts and tokens with Claude, then use Codex for robust implementation of the technical components in your codebase.
Implement a QA loop for AI-generated code using Devin. This playbook combines automated code reviews with an agent-recorded video walkthrough for visual verification, ensuring high-quality merges at scale.
Build a reusable 'Watchdog' playbook in Devin to automatically monitor customer accounts. The agent proactively checks for errors, synthesizes a prioritized list of critical issues, and saves you time.
Join 100,000+ product managers who use ChatPRD to write better docs, align teams faster, and build products users love.