Cordoba started in 2023 as an architecture AI created for testing and projects for Architects West in Coeur d’Alene, where our founder Morgan Dixon’s father, Keith Dixon, worked. He met with the team at Architects West and explained to them that he was working on an architecture AI and needed to train data. That’s where it all started.
In June 2023, throughout the summer, he worked on the architecture and platform, designing it and making it work as best he could for different iterations of the model architecture. The architecture data included full house designs, as well as codes and other data that is important for building a house, so that we could completely train a model to develop entirely completed houses that were engineering compliant as well.
From there, Morgan entered school at the University of Idaho, completing his senior year and degree from 2023 to 2024. Throughout the year, he met a few new people and competed in a few pitch competitions. During one of his competitions called Sparks Weekend in Spokane, he met Chris Larsen, who would become important in the story momentarily.
From there, the year continued until Christmas break, when Chris invited Morgan and one other friend to come and discuss plans for the next term’s business plan competitions across the nation. A few different ideas were discussed, but Morgan decided to bring up his architecture AI, which was at that point called Jupiter.
After this initial meeting, they determined that this was going to be the best project for the pitch competition season, so they continued to use it. Morgan developed the AI, and by late February, he had a fully working text-to-3D architecture AI built entirely from scratch with their own unique model architecture. It was able to train locally on a 2016 MacBook Pro with an i5 core and only required 32 MB of storage. This extraordinarily lightweight text-to-3D model was a state-of-the-art architecture.
From there, the rest is history. They won a few different pitch competitions at WSU, University of Idaho, and competed at the University of Washington and Boise State University, where they made it to finalist rounds. In total, during the competitions, they won over $19,000, and in the summer of 2024, they joined the University of Idaho’s 2024 ICORE program, which awarded another $3,000 for completing discovery customer calls and interviews.
At the end of the competition season and school year, Chris decided to move to Utah with his girlfriend and follow his religious goals to become closer with Jesus and the Church of Latter-day Saints. We are very thankful and proud of what Chris is doing and following his path with Jesus.
After Chris left, Morgan decided to focus on the gaming industry and convert Cordoba into a gaming platform for eventual text-to-3D video games. Having a model that could do architecture was cool, but to scale it up and make it super high-quality required a lot of computing power that the budget just didn’t make sense for.
While Cordoba was being produced, the model was also working on bringing in real-world data, which eventually turned into what it is now: pulling in real-world data for video games. We decided to go about this approach because it was simpler and required less computing power, so it would be faster to revenue versus approaching it as an AI model. We eventually want to return to the point of training a model based on all the generations and data they have collected so we can do text-to-3D worlds, which we will get to, but in the meantime, it is cost-saving and personal time-saving because Morgan is running the whole project. It is important to go after what is the fastest revenue.
The ideas for going after the gaming industry and Unreal Engine were also suggested by David Barr, a former WSU business plan competition winner, as well as Ryan Isbell of Intract. Both have been amazing mentors in guiding the goal and launch of the Cordoba plug-in for Unreal Engine.
As I am writing this post on September 2, 2024, the website for Cordoba and the API have been created, and we are just waiting on the plug-in to be completed and launched. From here, we will develop plug-ins for Unity, Roblox, Minecraft, and Rabbit, so that we can cover the bases of world-building and help get one step closer to text-to-video games, starting with the world.