Hey, so something weird happened to me yesterday. An agent appeared to have some existing memory about my objectives and what where doing, even though it was fresh one without any summary from previous chats and with very limited other context.
It’s kind of hard to explain. It was actually not a great memory as it was kind of out of date. But like, I’ve checked my documentation in the codebase and I don’t think it could have logically reached that assumption without some sort of memory storage.
Kind of interested to know peoples thoughts. I know there our products like https://mem0.ai/ out there because I tried Pear.AI during the 0.47 dark days - and it was interesting but probably useless your starting a codebase fresh from the init up.
Memory much?