This project is an experiment in building a production-ready game without writing code by hand.
The goal is to document the process of creating a fully functional, monetized, multiplayer game using only AI-generated code—primarily through Codex—while focusing human effort on architecture, decision-making, and iteration.
Software development is undergoing a fundamental shift driven by rapid advances in generative AI. This project explores a key question:
Can a complex, real-world product be built entirely through AI-generated code, with humans acting primarily as architects and reviewers?
To answer this, the project documents the end-to-end development of a Risk-style strategy game, capturing both the successes and the friction points of “vibe programming.”
- A case study in AI-driven software development
- A practical experiment in zero hand-written code
- A documented journey of building and shipping a real product
- A testing ground for AI-assisted architecture and iteration
- A tutorial on how to run the code
- A simplified demo or prototype
- A purely theoretical exploration
This is a production-focused build, with real deployment targets, real constraints, and real tradeoffs.
This project is designed for:
- Engineers interested in AI-assisted development but with limited hands-on experience
- Engineering managers evaluating how AI may impact team structure and productivity
- Technologists curious about the future role of programmers
The content will include:
- High-level blog posts for broader accessibility
- Detailed breakdowns of individual coding sessions for technical readers
The end goal is a fully functional, cross-platform strategy game inspired by classic Risk gameplay, including:
- Multiplayer gameplay with matchmaking and rankings
- Deployment to iOS, Android, and web (WASM)
- Monetization systems
- Community-driven content
- League and tournament systems with competitive play
Additionally, the project explores AI-driven player interaction, including systems where AI agents can communicate and form alliances with human players.
- No hand-written application code
- All code is generated using Codex
- Code is reviewed, guided, and iterated on—but not manually authored
- Development is performed in IntelliJ IDEA
- Running, debugging, and validation are human-driven
If a situation requires manual code intervention, it will be explicitly documented as a break from the constraint.
Generative AI is rapidly changing what is possible in software development.
This project investigates whether:
- AI can handle the full lifecycle of a complex system
- Developers can shift from implementers to architects
- High-quality products can be built faster with fewer traditional coding steps
The outcome is not assumed—the goal is to observe, measure, and learn.
The project is considered complete when:
- A fully playable game is deployed to:
- iOS and Android app stores
- Web via WASM-compatible browsers
- A multiplayer backend is live and supporting real users
- A series of blog posts documents the journey, decisions, and lessons learned
Success is defined by two things:
-
Delivery
- A working, production-ready multiplayer game
-
Insight
- Clear understanding of the strengths and limitations of AI-driven development
- Practical lessons on what accelerates or slows down “vibe programming”
The role of the programmer is changing.
What once required deep manual implementation is increasingly achievable through AI-assisted workflows. This project is an effort to engage directly with that shift—early, seriously, and in a real-world context.
The goal is not just to adapt, but to help define what effective development looks like in this new landscape.
- Build toward production, not prototypes
- Optimize for speed of iteration and learning
- Break work into incremental, shippable milestones
- Document what works—and what doesn’t
This repository is licensed under Business Source License 1.1 (BUSL-1.1).
- Source code is publicly visible for learning, review, and non-production use.
- Running a production or public hosted version is not permitted under this repository license.
- The license includes a far-future change date of 2100-01-01.
See LICENSE for the full terms.
If you want to support this project, you can contribute here: