"Together, memory and dreaming form a robust memory system for self-improving agents. Memory lets each agent capture what it learns as it works. Dreaming refines that memory between sessions, pulling shared learnings across agents and keeping it up-to-date."
This is really helpful for those large system's that need the long memory to ensure that you do not have to "reteach" the AI from a base memory or md file.
This I believe will go a long ways to improving token usage/caching and at the end of the day improving those large codebases.
Link to sign up to their managed agents platform.