I’m a marketing technologist, not a game developer. But I managed to ship a real-time cooperative multiplayer game built entirely with Claude Code. Here’s what I learned about product design, AI-assisted building, and why distribution is still the hardest problem.
When I was a kid, I spent an unreasonable amount of time playing a game called Bolo on the Apple II. It was simple, top-down tank game. My brother controlled the movement, and I would control the gun on the other side of the keyboard.
Fast forward 40 years to the Switch era, when I played Lovers in a Dangerous Spacetime with my kids. The concept was multiple players share a single spacecraft, and each player controls a different system. One person steers, another shoots, another manages the shield. You physically can’t do everything alone. The fun isn’t in individual skill, it’s in the chaos of trying to work together under pressure. Which was the precursor to games like Overcooked.
So I thought: what if you took that co-op concept and put it in the browser? No downloads, no installs. Just share a code with your friends and you’re in. An .io game, but one where teamwork isn’t optional, it’s the whole point.
So I built it. The game is called VoidHQ.io, and it’s live now.
What VoidHQ.io actually is
VoidHQ.io is a cooperative multiplayer space game where up to four players share control of a single spacecraft.
Each player picks one role:
The Pilot controls movement. The Left Gunner and Right Gunner each aim and fire an independent weapon. And the Warden controls a rotating shield that blocks incoming fire.
The shield is indestructible. But it only covers part of the hull. So the Warden has to constantly rotate it to face whatever’s shooting at you, while the Pilot maneuvers to give the Gunners better firing angles. If you and your crewmembers don’t work together, you blow up.
You start in a safe zone, mine asteroids for chemical elements, and use those elements to upgrade your ship’s systems. Each upgrade is a permanent, irreversible choice; do you want your weapons to fire faster or hit harder? Do you want the shield to cover more area or to reflect projectiles? These choices define what your ship becomes by the late game, and every crew ends up with a different build.
The entire thing was built with AI
The code for VoidHQ.io was written using Claude Code. I did not hand-write the game engine, the networking layer, or the rendering pipeline.
What I did write was the product requirements document. The gameplay mechanics. The tech trees. The upgrade economy. The enemy behaviours. The zone structure. The death and respawn loop. The lobby and matchmaking flow.
Once i had a working prototype, my kids jumped in with even more ideas, which I then instruct to Claude Code, and deploy it in minutes.
The downside is, they have become “that client” that asks for a feature and expects it to be done by that same evening.
So the code part was hands off. But the product work, and whether the game feels “fun”, is the part that AI can’t do for you. Yet.
Anyone can now prompt an AI to “build a game.” You’ll get something. But the decisions that you make are what creates something unique. Those are design decisions that come from years of playing games, years of building products, and a clear understanding of what makes systems feel good to use. That’s product management, not prompt engineering.
The hard part is not the build
This brings me to something that a lot of builders like me are realising right now: the build is getting easier, but the distribution is hard.
Tools like Claude Code have genuinely compressed the time it takes to go from idea to working product. The gap between “I have a spec” and “I have a thing people can use” has shrunk dramatically.
For me, it hasn’t just “compressed” the time, it has made it actually possible in the first place.
The last few months have been incredible for solo builders. I have shipped a payment plugin for WooCommerce, created AI content workflows, built a crisis simulation platform, and now a real-time multiplayer game. The range of what one person can build and launch has exploded.
But getting people to actually use the thing? That hasn’t changed at all. If anything, it’s harder, because the same tools that let me build a game also let a thousand other people build a thousand other things, and everyone is competing for the same attention.
For VoidHQ.io, this is the challenge I’m working through right now. A multiplayer game has a cold-start problem that most products don’t: it’s only fun if other people are playing at the same time. A landing page can drip traffic over weeks. A game needs enough concurrent players to fill crews, or the experience is sub-optimal.
So the product skills got me to a working game. Now the marketing skills need to get me to a playing population. That’s the real test.
What this means if you’re a product person
I think VoidHQ.io is a proof point for something that’s changing fast in the industry. The ability to implement software is becoming less of a bottleneck. What remains scarce is the ability to design systems that people want to use, make the right tradeoffs, and then distribute the result to the people who need it.
Those are product skills. Those are marketing skills. And for the first time, those skills alone without a team of engineers behind you, can take you from zero to a live, working product.
I’m not saying the engineering doesn’t matter. I still work with smart full stack devs in Sejasa.com and Recommend.my, but for this game, things like the binary WebSocket protocol, the client-side prediction, the physics engine, I didn’t need to be the one solving them. I needed to know what “good” feels like and be able to specify it clearly enough for the AI to execute.
If you’re a product manager, a marketer, or a founder with a technical vision but no engineering team, the barrier to entry just dropped by an order of magnitude. The question is no longer “can I build this?” It’s “what should I focus my attention on building, and can I get it in front of people?”
The first question is about product judgment. The second is about distribution. Neither one is new, and neither one has been automated fully.
Try it
VoidHQ.io is free, runs in your browser, and requires no download or account. Share a 4-character code with your friends and you’re in. If you don’t have friends available, you can still play solo. I made it so that empty crew slots run on a (dumb-ish) autopilot, so a human crew is still significantly better.
If you try it, I’d love to hear what you think. And if you’re working through similar challenges; shipping products with AI tools, solving the distribution puzzle, let’s talk.
Leave a Reply