Originally published as an Advisory Hour Substack post on 2026-08-27. This first-party site copy preserves the text, localizes public supporting media, and connects the record to its project trail.
Original post
Founderman2 is a reskin of the 1776 Megaman inspired game to test out an art production pipeline. The original game I jokingly called Founderman in honor of the playable characters who are the founding fathers of the United States. The original game was a full atari style game and it featured crisp gameplay that I just couldn’t shake off. I really enjoyed the jumps and sharp timing of the game, but felt it could be better.
During the creation of Mario Walker I realized I could use the new sprite techniques in other ways. A great test would be taking a game built over the summer and just retrofit the game using AI from this fall along with the new sprite based techniques I’ve developed. It wasn’t just any sprites-I wanted a new art style. A new kind of visual that no one had really seen in a megaman inspired game.
The stages are unique, the backgrounds have reasonable parallax, and on their own each screenshot reads very well. There’s always a lot of room for iteration and improvement, but given the release of transparent sprite support from the latest openai model it means that I can imagine entirely new types of games and just toss aside my Chroma Key pipeline. As the video shows, however, I did get pretty far with it.
The charge attacks, slide, special weapons, bosses and more are still in the game. The stages are much longer. The jumps are more “human friendly”, and there’s some surprise traps and more to learn about on each stage. From a gameplay perspective, while it wasn’t the main focus, it was fun to get the revamp done. This even includes some sprites for the special weapons available by default for each of the four characters, such as Ben Franklin’s Kite.
Founderman2 with its longer stages might trigger a memory of what I wrote about in Mario Walker about longer gameplay and stage development. These are no longer quick in and out stages. While the stages are not masterpieces, they are longer and more interesting. The jumps and challenges and learning loops do stack. There’s something resembling gameplay happening here. The big unlock for me to figure that out might surprise you: I just started reading game design books.
The more you can convey to the model to do on your behalf, the better your output in such a short time. Try it yourself using my website’s gamelab (https://ericrhea.com/gamelab) and asking an AI model to use the game design practices at work in gamelab in your next project. First, switch to plan mode, then reference my website and send the agent out to extract all the game design lessons I’ve hidden away on the builds. Then map those to your project. You might find it has some interesting results!
Here’s an example of what I mean by this on a brand new project, flip on plan mode:
And with any luck, you’ll get game design tricks packed and ready to go from the very first question:
Now to see where this “truck mud” project might end up at.
Where to play Founderman2
Play and learn more here: https://ericrhea.com/gamelab/founderman-2.html
The original and its pixelated glory still has a place in my heart, however.