Inspired by ReasoningBank research, create storage to keep memories about agent previous user interactions. This memories will include distilled summary of what went welll or bad and reasonning on how to improve future similar interactions. These memories can be loaded as initial Agent context based on problem statement. We expect Agent perform better based on previous experience: more successfull interations, less steps to achieve desired result.
Inspired by ReasoningBank research, create storage to keep memories about agent previous user interactions. This memories will include distilled summary of what went welll or bad and reasonning on how to improve future similar interactions. These memories can be loaded as initial Agent context based on problem statement. We expect Agent perform better based on previous experience: more successfull interations, less steps to achieve desired result.