We Made a Video Game: Here’s How We Did It – by BenjiPays
Conceptualizing the Game
When we started, we knew two things. One was that we wanted to make a video game. (You can read about the why further on). The other thing we knew is that it’s actually something anybody can do now. I think Claude Code had been out for a couple of weeks, and there was another tool called Rork that we’d been playing around with as well.
The reality was that even a few months ago, you couldn’t just make a video game. Or at least, you couldn’t do it easily.
The first and most obvious question when you sit down to make a video game is to decide what kind of game you want to make. We wanted to build something anyone could pick up in thirty seconds without downloading anything or creating an account. Just scan and play.
We landed on the idea of an endless runner type of game. Endless runners are great because of the loop they create. A session is short enough to finish in under two minutes. The format is immediately understandable with no instructions needed. And the score-based structure means the natural reaction when you finish is to want to play again. One more run. Just one more.
Building the Game
Once you know what you want to build, you fire up Claude Code, and you vibe code a prototype. That part took maybe 2-3 hours. We described the type of game, who the mascot would be (Benji, of course), and what Benji would be doing. Obviously, that would be chasing invoices.
Successive prompt iterations went into detail about the different levels, the speed of gameplay, and the visuals.
The prototype was surprisingly viable, and that allowed us to move into the next phase of development by the second day. We did this by splitting into two teams. One team focused on the visual aspects of the game. We took images of cityscapes and asked Claude Code to redesign those images to fit the game’s aesthetic.
We did the same thing with the background music. We knew from the start that music couldn’t be an afterthought. In most mobile games, it’s background noise. We wanted it to be part of the gameplay, something that changed the feeling of what you were doing in real time. So we built it that way. Each city has its own soundtrack pulled from regional musical influences that feel true to that place. When you hit a power-up, the game shifts from day to night and the music shifts with it. The tempo picks up. The energy changes. You feel it.
The other team focused on aspects of the gameplay. We added power ups, for example. But we also leaded things like a leaderboard. The leaderboard is an important part of gameplay – it takes you back to the arcade, when seeing your name on the game’s screen was all the motivation you needed. And seeing someone else take over your position would send you to the bank for a roll of quarters.
The last step was adding in things like security, hosting and the rest of the back end.
The Hack for Improving a Video Game
We’ve been taking the game around to a few shows and we’ve got quite a bit of feedback on it. That feedback has been used to iterate on the game. We’ve made some serious improvements to the gameplay, and the visuals, and it’s quite eye-popping now. Especially compared with how rough those first versions were. Testing on multiple devices and with different users helped make the game better.
The real hack, however, was using our kids as beta-testers. It was obvious pretty quickly that they were the most enthusiastic, played a lot, and gave us great feedback. They also set all the early high scores, as is the way with video games.
Why a Video Game?
Video games are a different way of telling stories. In the MSP world, we all grew up on video games, and it’s just something unique that helps to tell our company story. Ok, the game makes chasing invoices sound more fun than it is in real life, but the point is the same. Benji does it.
Of course, the other reason is because we could. There’s no way we would have been able to do this before Claude Code, and we were pretty quick to jump on the new technology when it became obvious that it would allow us to build a game quickly.
Final Thoughts
What’s really cool about working with the AI tools know is that building a video game is mostly about the concept. We got together as a team and thought through the different elements of the game. We focused on the experience. That’s because the rest of it is basically vibe coded.
It took a few hours of brainstorming before we started prompting. It then took a few more hours to prototype.
It was another couple of weeks before we finished the 1.0 version that we released to the public. That was because we had to play the game over and over to test individual elements that we could refine. Literally a small team of people with no prior experience building a video game just vibe coded a really cool game in a couple of weeks, and now you get to be the beneficiary by having a little fun, and talking to the lead creator Rafael at the show about how he did it.
What that means is that once you have an idea, you can make a game of your own, with no code, just ideas about the type of game you most want to play.
You can play our game here: benjigame.com or drop by the Benji Pays booth to give it a shot!
We’re ever offering prizes for the high score – but the real prize is the prestige of winning. Or, if you like, picking our brains about how to get started building a game of your own.
