MemGPT

An agent design in which a language modellarge language model, n.: A computer program trained on enormous amounts of text to predict the words that come next. At vast scale, that simple skill lets it chat, answer questions, translate, summarize and write code. curates its own memory. Only a small working set fits in its context windowcontext window, n.: How much an AI model can read and keep in view at once: the instructions, the conversation so far and any attached documents. Text that does not fit has to be cut, summarized or left out., so the model is given functions to search, insert, edit and summarize material kept outside — a searchable log of past messages and an archival database — and decides for itself what to bring in or move out, prompted by system messages such as a warning that the window is nearly full. Its authors called the scheme virtual context management, by analogy with an operating system shuttling data between main memory and disk, and showed it working through documents too large for the window and keeping track of facts about users across many chat sessions. MemGPT was an early example of self-editing agent memoryagent memory, n.: The ways an AI assistant or agent keeps information — your preferences, past conversations, lessons from earlier tasks — and brings it back later, beyond what fits in a single conversation. and of the compactioncompaction, n.: Like replacing a stack of rough notes with a one-page digest: when an AI's conversation grows too long for its memory, the history is condensed into a summary and the work continues from there. and offloading techniques later grouped under context engineeringcontext engineering, n.: Managing an AI model's reading material: choosing, trimming and arranging the instructions, documents, tool results and memories it is given on every turn, so that it has what the task needs and little else..

MemGPT-style agents decide for themselves when to archive a fact and when to search for it.