FIELD NOTES
Continuity is not the same as learning.
Keeping history is different from improving the next cycle.
Most teams now save AI outputs, prompts, and previous versions. That is memory, not learning. Memory is a record of what happened. Learning is what makes the next cycle better.
Learning requires the system to use history: to remember what was rejected and why, to carry feedback forward, to reduce repeated explanation, to improve context over time. Without that, every session starts from zero.
Memory is a record. Learning is a better next step.
The test is simple. If the next brief, draft, or campaign begins in the same place as the last one, you have continuity without learning. If it begins one step ahead, you are learning.
The goal is not to store more. It is to make the stored information useful for what comes next.