I continued to upgrade the existing autobot until it became a multi-autobot, but the efficiency was not very good, so I finally agreed with the agent to remake it from scratch.
So, for the new autobot, we decided to make a kernel first like Linux and then build on it. We are currently training the core.
I asked Jemina to create a diagram of the approximate structure.
The text is a bit broken, and it feels crammed, so I asked Claude Design to create the same content in a diagram.
This is extremely simple.
For now, we are training the core and plan to attach adapters later to create genre-specific game adapters.
The agent says it's better, but in reality, the existing standalone autobot learns faster. The new version is structured so that its ability increases proportionally to the amount it learns, so a lot of basic learning is required.
After game learning is complete, I'm tired of uploading new games and asking Aiden to change them every time. This needs to be automated.
