$ ./game --agents▊
Stop at the concept stage no more. Learn to design, direct and deploy multiple open-source AI agents that generate gameplay, build worlds, write narrative, test balance and fix bugs alongside you — turning your game idea into a finished, playable, production-ready video game. Run every model on API keys with a free daily tier.
Most devs never get past a design doc or a grey-box prototype. This course exists because of a gap I kept seeing: nobody teaches you how to build a complete, playable video game — efficiently — by turning a crew of open-source AI agents into your dev team.
You bring the game idea. I bring the architecture — how to orchestrate agents that generate, code, test and critique one another, then connect their output to a real game engine. Together we ship a full, playable game in production — not another prototype graveyard entry.
Full loop: world, gameplay, levels, narrative, audio and tests — playable end-to-end and ready to put in your portfolio as a shipped product.
Designer, level-builder, narrative lead, art asset, QA and balance-test agents — coordinated crews that generate, critique and improve each other's work while you stay in control.
All LLMs run through API keys from providers with a free daily tier — build, iterate and even playtest before spending a cent on tokens.
A complete game touches many disciplines. So your agent crew has many seats. Each agent is a specialist, and together they work like a studio — through a shared pipeline you orchestrate.
Turn a rough idea into a scoped Game Design Document, stress-test it, and design the rules, progression and core loop. These agents keep your design coherent as it scales.
Generate coherent lore, dialogue, quests and levels with RAG pipelines — content that stays consistent with your world, tone and style guide, adapting live to the player.
A style-guide agent that keeps every generated asset on-brief — from 3D-ready references in Maya+Pipeline to textures, sprites and UI — preserving your game's aesthetic while scaling production.
Autonomous playtesting crews that hunt bugs, exploit unintended outcomes and sim thousand of runs to rebalance win rates — so your data, not your gut, drives the final polish.
Fourteen sessions, seven weeks. Every topic is taught hands-on — you build as you learn, and each session builds toward your final playable game. Written in plain language that anyone can follow.
See how the course is structured and how each session builds toward your final game. Meet your toolchain, learn who your agent crew is, and get your environment ready — agents, engines, API keys and all.
Understand the shift from hand-coded systems to LLM-powered agents that infer and act. Learn how different agent architectures shape behavior, the difference between agents used as dev tools and agents embedded in gameplay, and the "AI contract" that keeps everything predictable.
Translate your game idea into a Game Design Document that technically aware. Discover how AI changes design decisions (token budgets, agent roles), scope for creativity AND feasibility, and pick the "one agent, one wow" focus that keeps your scope from exploding.
Before writing real code, use agents to simulate players: spot gaps, friction points and dead ends in your design. Learn to iterate on structured feedback from both AI and human collaborators so you build the right thing.
Design a hierarchical system of agents that produce game-ready output together. Learn task delegation, manager agents, shared memory, and how to hook everything into your game engine and file system — then document the architecture.
Use retrieval-augmented generation so your agents produce dialogue, quests and assets that stay consistent with your lore. Control persona and tone, and run automated consistency checks that keep quality high at scale.
Turn raw agent output into structured formats your game engine reads. Learn import automation, how to write scripts that bridge agent output to gameplay, and watch a live terminal drive a game engine as content flows in.
Go beyond content generation to agents that actively participate in gameplay. Build goal-oriented action planning, utility scoring, perception systems and memory decay so your agents feel responsive and believable inside the game itself.
Apply agent loops to level design. Agents generate layouts, test them against gameplay constraints, and refine them — producing creative yet navigable, playable spaces. You'll build a dungeon-builder system that keeps worlds coherent even at scale.
Explore the difference between human-made and AI-generated assets, and how to preserve your artistic direction inside automated pipelines. Build a style-guide agent that keeps every generated asset on-brief, and talk honestly about ethics and audience.
Engineer narrative systems that react in real time. Track story state, keep characters consistent with dedicated narrative-agent, and generate dynamic dialogue that responds to what the player actually does — not just what you scripted.
Train adversarial agents that deliberately hunt bugs, balance problems and unintended outcomes. Simulate thousands of runs, analyze the results, and use the data to polish your design and stabilize your game.
Connect every agent & pipeline from previous weeks into one "shipping build". Balance performance, cost and quality, choose between local and cloud models, and run a cost analysis so your game doesn't bankrupt the developer behind it.
Do the technical triage: optimize your agents, compress prompts, review dev logs and polish performance and clarity. Then present your functional prototype to peers, focusing on live agent behavior and log transparency.
Explore the ethical and practical side of deploying agent-driven games — guardrails, player safety, and the future of AI in game engines — and learn how to present your work to players and recruiters with confidence.
Every session builds toward one thing: a complete, playable game powered by your own multi-agent pipeline. Your agents become your dev team — generating content, testing gameplay, and driving production end-to-end. You graduate with a portfolio piece you can actually play.
You don't need a big ML budget to build a game with agents. All models in this course run through API keys from providers that include a daily free tier — so you build, iterate and ship your first playable build without paying for tokens.
The provider-agnostic stack behind your dev agents — plug free-tier models directly and route across providers from one key system.
Gemini's free daily tier gives generous quotas for text, vision and voice — ideal for narrative, art direction and windows lean studio builds.
One key, hundreds of models — including free models with daily limits. The ultimate routing layer: pick the best model for each agent without changing code.
Throughout the course you'll wire real API keys from these providers into your agent crew — and learn cost tracking, rate-limit handling, retries and model selection. Exactly the skills you need to keep an indie game production affordable and profitable.
If you've got a game idea that keeps dying at the concept stage — and you're serious about shipping — this course was built for you.
You've got the ideas but not the time or team to build at scale. Assemble an AI dev crew that generates, tests and refines, while you stay in control.
Move past demos to production systems — orchestrate multi-agent pipelines that plug into real game engines, automate testing, and keep costs in check.
Stop fighting to keep systems consistent. Build AI-driven tools — level generators, style-guide agents and narrative engines — that scale your ideas without breaking them.
14 sessions · 7 weeks · playable capstone
for teams shipping games with agents
Indie studios and small teams shipping games with a multi-agent AI workflow get a tailored track with shared architecture review and delivery support.
"Not tuition — the fastest route from idea to a playable game."
Book a 20-minute call. Tell me the game you want to build — I'll tell you the agent crew, the architecture and the path to a playable build. No pitch, just game engineering.
I'm a computer science engineer with 8+ years shipping software across health, finance, telecom and logistics. I grew up on Unity 4 & 5 and Maya with Python, and throughout 2026 I've been building GenAI applications at etheroi.com using open-source agentic projects — the same engineers I'll teach you how to orchestrate for your game.
No fluff, no throwaway prototypes. I bring the same production-first rigor I use to ship real software and apply it to your game — so you graduate with something you can actually play, not something you demonstrate.