Like many great inventions, Bitmagic happened almost accidentally. We were building something else - a role playing game where one of the players would be the game master - and we needed a tool that would allow creating and modifying game worlds really quickly based on instructions given in natural language.
We had a prototype of this tool ready in early 2023. To me, this started to feel like something much, much bigger than a tool for game masters. I believed that the power of generative AI would allow building entire games quickly and easily. It would allow creative people to design and build games on their own. In fact, it would allow anyone to create games.
The very first prototype of Bitmagic (back then still known as Roleverse)
I’ve been in the games industry for 30 years. I made my first games as a teenager, and I’ve worked on various roles over the years - I was one of the first employees at legendary indie studios Remedy and Housemarque, and I was a lead programmer at Westwood Studios working on the Command & Conquer franchise. Since then I’ve been a CTO, CEO, programmer, designer, you name it. But one thing has always been clear - making games is extremely difficult and requires the kinds of skills and experience that has always made game development an art form of the select few.
Even today, billions of people play games, but all of the games out there are made by just some hundreds of thousands of people - mostly white, middle aged men. Suddenly we have a chance to democratize game development and open the doors for millions more game developers who had probably never thought they could make games!
A decision was made in the spring of 2023 to pivot the company. We scratched the plans to create the role playing game and focused our efforts on one thing only - building a platform that allows anyone to build games of their dreams. The platform was named Bitmagic, because it really feels like magic when you describe a game you’d like to build, and moments later you are playing it.
Adding demolition physics to Bitmagic game engine
In May, we set out an impossible goal - we would launch Bitmagic by the end of November.
This meant taking an early prototype and turning that into a fully featured product in less than six months! In the end, we made it, and Bitmagic Alpha was launched on the last day of November. Of course, it’s not quite a fully featured product. Bitmagic is an incredibly complex system that incorporates generative AI and a game engine in a way that has never been attempted before, so it takes time to get it right. What we have now is an alpha version of something that will be fully completed over the next 12 months, but nevertheless - it’s already being used by people from around the world. And we’re hearing a lot of praise and excitement.
The current version doesn’t allow users to save their creations or share their products, and the number of people we have allowed on the platform is limited… but soon all of that will change.
Back to the demo scene, mentally
Bitmagic has been the most fun I’ve had coding since the demo scene days in the early 90s. We’re building something that hasn’t been done before, nobody really knows what is going to be possible with this thing. It feels like we’re just scratching the surface, as we keep inventing new ways to use generative AI as we go along. Unlike AI models that create images or video, a generative system that creates games cannot be made with a single AI model. The stack of systems is complex, and each component has to work independently, but still produce something that works together with everything else.
Some of the things we create on the fly: the world and a background story, design for the game, NPCs, weapons, 3D objects, behavior code… everything. When the application starts, we have an empty scene, with nothing in it. As an example, when the system creates an NPC, it might also create a weapon for the NPC to use - since we have no idea what the NPC will be like, nor the weapon, our system needs to be able to write code on the fly that gives the NPC the ability to use the weapon that was just created. When the system creates code, the code might not work, so it needs to be able to detect when code doesn’t work, debug it and fix it. While this sounds quite hard, we have an incredible team of really smart fearless developers, and so far we’ve been able to do everything we decided to do, on schedule. This work is fun and rewarding, but certainly not easy.
"Add a shotgun that shoots exploding cheese"
What’s interesting is that even though we get a lot of help from the AI, the need for some serious hardcore coding is ever-increasing. It seems the more we do, the more there is to be done, so the need for coders is not disappearing. At least not in the near future. The theme for 2023 was that thanks to coding helpers like Copilot and ChatGPT, everyone is able to do more than ever before. But rather than lessen the workload, this has enabled us to build things we could not have believed we could do in the time frame and with the team size. My prediction is that this will continue at least for some time - the need for skilled programmers who know how to utilize AI in their work will greatly increase.
As a side-note: The world will likely see big layoffs in 2024 across many industries, as AI adoption means productivity per person goes up, but I believe this will be a temporary shift and new jobs will be created once businesses start realizing that while they can do more with fewer people, the total expectation of what will be achieved goes up at the same time. There’s no way to stop this progress, so the best an individual can do right now is to make themselves ready for the new era where AI is part of our everyday work.
Going forward, in 2024
During the holidays I took some time to reflect what we achieved in 2023 and what we can learn from the past year. I wrote a plan of action for 2024, so we know exactly what will happen this year. I know what will be in all the milestones, and the dates they will be completed. I even know when we’ll be raising money, and how much.
However, the future is always hard to predict, and it’s getting even harder now that AI moves forward at an ever-faster pace. What was considered impossible, is now possible. And what seems like a daydream now could be reality in 6 months. So, my carefully laid out plan will likely be rewritten many times this year. Even so, writing out detailed plans for the near-term future is useful because it helps you understand what needs to happen for you to reach your goals. My plan for the year has some black boxes - components that we will need that don’t exist yet, but that I know will be available this year. We don’t yet know who will make them, or any details really, but I know enough about what is happening in the AI industry that ’m 100% confident that they will be available when we need them.
There’s also the reality of being a small startup. We have a really amazing team, but I have things in my mind I would do right away if we had just some more funding so that we could hire a couple more people. VC funding rounds are slow and require a lot of work, so I’m needing to look for AI savvy angel investors willing to jump in between rounds. This means networking, attending events, speaking at conferences (such as CES), etc., which is important but it’s hard to find the time when the product needs attention as well. Building a startup is a delicate balance of being optimistic and always being on the edge, and making sure you stay alive for long enough to reach the next level!
All the best for 2024 to all startups out there, especially in the AI space! It's going to be an exciting year! Please reach out to me on LinkedIn. :)
The official alpha version trailer from November 2023
Following are some user-posted screenshots from our Discord (join today at https://discord.gg/bitmagic)