I Made a Game Jam Game Without Writing a Single Line of Code (Radar Blind)

If you’ve ever done a game jam, you know the drill: you consume an unhealthy amount of caffeine, stare at your IDE until your eyes cross, and furiously mash out code hoping the physics engine doesn't completely break at 3 AM. But a couple of weeks ago, I stumbled upon a jam hosted by a popular YouTube creator, and I decided to completely flip the script.

Working on my new game jam entry. This is the intro that I hacked so far. #gamejam #indiedev

[image or embed]

— kristijandraca (@kristijandraca.com) June 22, 2026 at 11:17 AM

The event was The Very Serious JuniperDev Game Jam, and I decided to enter a little project I call Radar Blind.

Here is the kicker: I made a hard rule for myself before the clock started. I would not write a single line of code. Not one. Every script, every variable, every logic loop was going to be generated by AI.

I even put myself in the game's credits as the "Executive Supervisor of Artificial Labor." Let me tell you, it was an absolute blast.

Shifting the Developer Mindset

Normally, my game dev workflow involves diving straight into the engine, writing some messy prototype code, and refactoring it later (or, let's be real, never). But when you rely entirely on AI to write the logic, you can't just wing it. If you feed AI vague ideas, it's going to spit out unplayable spaghetti code.

This challenge forced me to take a massive step back and focus entirely on system architecture and design. Instead of typing syntax, my job was to build a bulletproof blueprint. I had to understand the "why" and "how" of the game's mechanics well enough to explain them to a machine.

While the AI handled the heavy lifting in the code editor, I kept my hands dirty where it mattered most for the vibes: the 3D models and the sound effects. I wanted the game to still feel like my creation, with that distinct indie grit, rather than a sterile tech demo.

The Artificial Workflow

So, how do you actually build a game this way without losing your mind? It basically came down to establishing a really tight feedback loop. Here is how my "supervisor" workflow shook out:

• The Master Plan: I sat down and wrote a highly detailed design document. I outlined the core loop, the inputs, the physics expectations, and the exact state machines needed.
• The Project Manager: I fed this massive plan into the AI. To my surprise, the AI didn't just write code; it acted like a senior dev, splitting the jam into logical, bite-sized chunks that we could tackle one by one.
• The Playtest Loop: The AI would spit out a script, I would drop it into the engine, and then I’d playtest it.
• The Friction: When something broke—and trust me, things broke—I wouldn't fix the code myself. I would go back to the prompt, explain exactly what the bug was, how the game should feel, and let the AI rewrite it.

This loop was incredibly rewarding. It removed the minor friction of hunting down missing semicolons or debugging simple syntax errors, allowing me to focus entirely on game feel and project momentum.

Prototyping at Warp Speed (and Why Engineers Aren't Going Anywhere)

Let’s address the elephant in the room: Is AI going to replace game engineers? In my opinion, absolutely not. Writing code is only a fraction of what makes a game work. What AI does do, though, is blow the doors wide open for prototyping.

Instead of spending days laying down boilerplate physics scripts just to see if a mechanic is actually fun, I could test an idea in minutes. You get immediate answers to the ultimate game dev question: Does this actually feel good to play?

By leaning into this constraint, I ended up learning a ton:

• Mastering the Tools: I got a crash course in how to actually collaborate with AI, learning how to write sharper, highly specific prompts that got me exactly what I needed without the hallucinations.
• The Art of the Polish: Because I wasn't bogged down in syntax errors, I had the mental bandwidth to focus on what actually makes a game jam submission stand out: the polish. I could spend my time tweaking the sound effects, adjusting the 3D assets, and smoothing out the game loop right when it mattered most at the finish line.

Conclusion.

At the end of the day, Radar Blind is proof that the role of the game developer is evolving. We are moving from just being code-monkeys to being true directors and system architects. The tools might be getting smarter, but they still need someone at the helm who understands what makes a game actually fun to play. If you've been feeling bogged down by the sheer mechanical effort of side projects, I highly recommend trying a constraint like this. Grab an AI, give yourself a ridiculous job title, and see what you can build when you focus entirely on the big picture. Cheers to the next jam!