A working definition
AI-native game development is a production method, not a marketing slogan. In this method, language models and related tools draft, revise, and search large parts of the implementation. A human still decides what the game is, which system owns which state, what “done” means, and whether a build is allowed to ship.
That split matters. If you treat the model as the designer, you get a pile of features that compile and a game that does not hold together. If you treat the model as an intern who can type very fast, you still have to be the person who understands loops, feedback, economy, camera, save data, and the player’s actual minute-to-minute experience.
Martinez AI Studios Academy uses one sentence as a contract: AI can write the code. You still need to understand the game. The flagship program — AI-Native Game Development Mastery — is built around that contract. 170 bilingual lessons are free to read on this site (five stages, seventy-eight modules). LMS enrollment stays closed.
What it is not
It is not a promise that AI does everything. It is not a no-code product. It is not a claim that syntax is obsolete, or that traditional programmers are the enemy. Syntax can be generated. Judgment about systems cannot be outsourced without someone catching the lie.
It is also not a university degree, an accredited certification, or a job-placement program. The Academy is a private education product of Martinez AI Studios. When certificates exist after real completion, they are private certificates of completion. The flagship currently has zero published lessons, so it is not completable and not certificate-eligible.
We do not invent student counts, salaries, or Steam-success rates. CONTRABAND: Edge of the Fold is a real commercial title on Steam, created by Dero Lavigne and published by Martinez AI Studios. That fact supports teaching from a live process. It does not support fairy tales about overnight careers.
Why the Academy exists
The bottleneck moved. Generating a function is cheap. Owning a game is not. A model will happily invent a second writer for the same piece of state, skip a regression, or “fix” a UI bug by disabling the input the rest of the game depends on. Those are not theoretical risks. They are the class of mistakes verified CONTRABAND case studies document when an incident can be evidenced.
The Academy therefore teaches design judgment, system ownership, debugging method, QA, and release practice — the work that remains after the code appears. Dero Lavigne is the instructor and the developer of that case-study game. Martinez AI Studios is the publisher and the owner of the education platform. Those roles stay distinct on purpose.
What you still have to understand
A game is a loop with feedback. Input, simulation, and presentation have to agree. Economy and progression are systems, not item lists. Difficulty is more than scaling encounter numbers. Saves, builds, and store pages are part of the product, not a ceremony after “the fun part.”
You can learn those ideas without memorizing a language’s trivia. You cannot skip them and still direct an AI partner. If you cannot say which module owns a piece of state, you cannot write a safe prompt. If you cannot describe the player impact of a bug, you will accept a patch that only hides the symptom.
How this maps to the flagship
The public flagship is a professional program: five stages, seventy-eight modules, and 170 bilingual lessons free to read on this site. Optional Stripe support does not gate the text. LMS accounts, downloads, and certificates stay closed.
What you can study today is the full lesson path, the philosophy on this page, the systems essay that follows, the root-cause method, the verified CONTRABAND incidents, and the instructor context.