Training your personal AI allows you to create or update your personal AI model with your latest Memory Stack. Without training, the AI model is generic and is not a reflection of you.
Your personal AI is built to synchronously learn new data and asynchronously retrain itself. This means that you sync new Memory Blocks as you experience them, but your personal AI does not retrain itself every time.
It is expensive to do it that frequently and 1 new Memory Block has little effect on your personal AI. Instead, retraining your personal AI is most useful after you have synced a significant amount of Memory Blocks (e.g., after a trip, after a week of work, with a new report).
For those reasons, you control when you train your personal AI to create your first AI version. Before that, you cannot Message or Compose with your personal AI.
Retraining updates your personal AI's model with your latest Memory Blocks. This new personal AI model replaces your previous, as only one is kept active.
To read more about training your personal AI, read our detailed guide.