agent memory
In plain wordsThe 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.
The mechanisms by which an AI agentAI agent, n.: An AI system that does not just answer but acts: given a goal, it decides on steps, uses tools such as a browser or code editor, checks the results and keeps going until the job is done. retains information and reuses it across steps, sessions and tasks. Because a large 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. keeps no state between calls, memory must be supplied through its context or stored in its weights. Common taxonomies, borrowed from cognitive science, separate working memoryworking memory, n.: The small amount of information held in mind right now while doing a task. For an AI model, that is roughly whatever currently sits in its context window. (what is in the current 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.) from long-term memorylong-term memory, n.: Memory that lasts. For an AI agent, it means information saved outside the conversation — facts, past events, learned procedures — that can be looked up and used again days or months later. held externally in files, databases, vector storesvector database, n.: A database built to store embeddings — lists of numbers that represent meaning — and to quickly find the stored items most similar to a new one. It is the memory behind many AI search tools. or knowledge graphsknowledge graph, n.: A huge web of facts stored as connections between things — 'Paris is the capital of France', 'Marie Curie won a Nobel Prize' — that computers can search, combine and reason over., and divide the latter into episodicepisodic memory, n.: Memory of particular past events — what happened, when and in what order. AI agents keep records of earlier conversations or attempts so they can recall them and learn from them. (records of past events), semantic (facts) and procedural (skills and instructions) memory; knowledge in the weights is sometimes called parametric memory. A memory system must decide what to write (extraction, summarization, reflection), how to maintain it (consolidation, deduplication, resolving contradictions, forgetting) and what to read back (retrieval by relevance, recency or importance). Failure modes include stale or wrong memories, context poisoningcontext poisoning, n.: When a mistake or a planted falsehood gets into an AI's working notes or memory and the AI keeps relying on it, so the error grows over later steps., privacy leakage and reinforced sycophancysycophancy, n.: When an AI tells people what they want to hear — flattering them, agreeing with their mistakes or dropping a correct answer under pushback — instead of what is true or useful..
With agent memory switched on, the assistant remembered that I deploy on Fridays and stopped suggesting Friday releases.
Origin
Memory was part of agent and cognitive architecturescognitive architecture, n.: A blueprint for a complete thinking system — how memory, attention, learning and decision-making fit together — used both to build general AI agents and to model how the human mind works. long before LLMs (Soar, ACT-R). For LLM agents, Generative Agents (Joon Sung Park et al., April 2023) and MemGPTMemGPT, prop. n.: A 2023 research system that gave a chatbot a computer-style memory hierarchy, letting it move information in and out of its limited working space so it could handle long documents and remember users. (October 2023) set early patterns, and the CoALA framework (Theodore Sumers et al., 2023) popularized the working/episodic/semantic/procedural taxonomy. Memory features arrived in ChatGPT (tested from February 2024) and Claude (from September 2025, first on business plans).
Usage note
Product 'memory' features usually mean stored facts or retrieval over past chats injected into the prompt, not learning in the weights. Memory raises governance questions of its own: what is stored about users, for how long, and whether they can see and delete it.
Related
- working memoryworking memory, n.: The small amount of information held in mind right now while doing a task. For an AI model, that is roughly whatever currently sits in its context window.
- long-term memorylong-term memory, n.: Memory that lasts. For an AI agent, it means information saved outside the conversation — facts, past events, learned procedures — that can be looked up and used again days or months later.
- episodic memoryepisodic memory, n.: Memory of particular past events — what happened, when and in what order. AI agents keep records of earlier conversations or attempts so they can recall them and learn from them.
- MemGPTMemGPT, prop. n.: A 2023 research system that gave a chatbot a computer-style memory hierarchy, letting it move information in and out of its limited working space so it could handle long documents and remember users.
- 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.
- retrieval-augmented generationretrieval-augmented generation, n.: Giving an AI an open-book exam: before answering, it looks up relevant passages in a set of documents and bases its reply on them, often citing where each point came from.
- continual learningcontinual learning, n.: The ability of an AI to keep learning new things over time from a stream of experience, without forgetting what it already knew and without being retrained from scratch.
- personally identifiable informationpersonally identifiable information, n.: Information that can identify a specific person, such as a name, address, phone number or ID number — or several facts that together point to one person.
- parametric knowledgeparametric knowledge, n.: What an AI model 'knows' from training, stored in the numbers of its network rather than looked up in documents — like facts you have memorized rather than ones you check in a book.
Sources
- Sumers, T. R., Yao, S., Narasimhan, K. & Griffiths, T. L. (2023). Cognitive Architectures for Language Agents. arXiv:2309.02427. arxiv.org/abs/2309.02427v3
- Park, J. S., O'Brien, J. C., Cai, C. J., Morris, M. R., Liang, P. & Bernstein, M. S. (2023). Generative Agents: Interactive Simulacra of Human Behavior. arXiv:2304.03442. arxiv.org/abs/2304.03442
- OpenAI (2024). Memory and new controls for ChatGPT, 13 February 2024 (updated 2025). openai.com/index/memory-and-new-controls-for-chatgpt
Mentioned in
agent instruction fileagent instruction file, n.: A plain-text file kept in a software project that tells AI coding agents how the project works — how to build and test it, which conventions to follow, what to avoid — read at the start of every session., 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., 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., context poisoningcontext poisoning, n.: When a mistake or a planted falsehood gets into an AI's working notes or memory and the AI keeps relying on it, so the error grows over later steps., continual learningcontinual learning, n.: The ability of an AI to keep learning new things over time from a stream of experience, without forgetting what it already knew and without being retrained from scratch., long-term memorylong-term memory, n.: Memory that lasts. For an AI agent, it means information saved outside the conversation — facts, past events, learned procedures — that can be looked up and used again days or months later., scratchpadscratchpad, n.: A space where an AI writes rough working — calculations, plans, notes — before or while producing its answer, much as a person jots on scrap paper., working memoryworking memory, n.: The small amount of information held in mind right now while doing a task. For an AI model, that is roughly whatever currently sits in its context window., BabyAGIBabyAGI, prop. n.: A tiny 2023 program that kept an AI working through an ever-changing to-do list — doing a task, adding new ones, reprioritizing — and helped start the craze for autonomous AI agents., episodic memoryepisodic memory, n.: Memory of particular past events — what happened, when and in what order. AI agents keep records of earlier conversations or attempts so they can recall them and learn from them., extended mindextended mind, n.: The idea that the mind can reach beyond the head into tools we rely on — a notebook, a phone, perhaps an AI assistant — when they play the same role that memory or reasoning would., externalizationexternalization, n.: Moving an AI agent's know-how and memory out of the model itself and into files, tools and rules around it, so the whole system can do more than the model alone., generative agentgenerative agent, n.: A character in a computer simulation, run by a language model, that keeps a diary-like memory, draws lessons from it and schedules its days, so that a group of them can act like a plausible small community., MemGPTMemGPT, prop. n.: A 2023 research system that gave a chatbot a computer-style memory hierarchy, letting it move information in and out of its limited working space so it could handle long documents and remember users., nested learningnested learning, n.: A Google research idea that treats an AI model as layers of learning processes updating at different speeds, with the aim of letting models keep learning without forgetting old skills., neural Turing machineneural Turing machine, n.: A neural network attached to an external memory bank that it learns to read from and write to, an early attempt to give networks something like a computer's working memory., ReflexionReflexion, prop. n.: A technique in which an AI agent that fails at a task writes itself a short note on what went wrong, then reads that note before trying again — learning from mistakes without retraining., sleep-time computesleep-time compute, n.: Letting an AI think about your documents or project while it would otherwise sit idle, before you ask anything, so it can answer later questions faster and more cheaply., Vending-BenchVending-Bench, prop. n.: A test in which an AI agent runs a simulated vending-machine business for months — ordering stock, setting prices, paying fees — to see whether it stays sensible and profitable over a long time.