Glossary20 terms
AI memory, defined in plain language.
An AI memory layer is software that stores what an AI tool should remember between sessions and returns the relevant parts on demand. The terms below cover how that works: context windows, compaction, checkpoints, recall, embeddings, MCP (Model Context Protocol) and the rest. Each definition stands alone. Each page notes how Central Brain handles the concept.
- AI memory layerAn AI memory layer is a component that stores information between sessions and returns it to a model, assistant or agent on demand.
- Agent memoryAgent memory is the stored state an AI agent carries between steps or sessions so it can continue work without being told everything again.
- Second brain for AIA second brain for AI is a personal knowledge store that you and your AI tools can both read, keeping context outside any single chat.
- Persistent memoryPersistent memory is information an AI system keeps after a session ends and can use in later sessions without being re-supplied.
- Context windowThe context window is the maximum amount of text, measured in tokens, that a language model can read and consider in one request.
- Context compactionContext compaction is the process of summarizing or trimming older parts of a conversation so a model can keep working within its context window.
- Memory checkpointA memory checkpoint is a deliberate save of an AI session's current state (decisions, facts, open questions, todos) to durable storage.
- Memory recallMemory recall is the step where an AI system searches its stored memory for relevant entries before answering a question or starting work.
- Memory compactionMemory compaction is the consolidation of many small memory entries into a smaller, current set, removing duplicates and superseded items.
- Semantic searchSemantic search finds text by meaning rather than exact words, by comparing numeric embeddings of the query and the stored content.
- EmbeddingsEmbeddings are fixed-length lists of numbers that represent the meaning of a piece of text, so similar texts have similar vectors.
- Vector storeA vector store is a database that holds embeddings and returns the nearest ones to a query vector, usually alongside the original text and metadata.
- Retrieval-augmented generationRetrieval-augmented generation (RAG) is a method where a system retrieves relevant documents first, then gives them to a language model to answer from.
- MCP memory serverAn MCP memory server is a program that exposes save and recall memory operations to AI clients through the Model Context Protocol.
- Model Context ProtocolThe Model Context Protocol (MCP) is an open standard for connecting AI applications to external data sources, tools and workflows.
- Source-traceable retrievalSource-traceable retrieval returns every search hit with the metadata needed to find it in the original document: file path, page, position.
- Local-first AILocal-first AI runs models and stores data on the user's own device by default, using cloud services only when the user opts in.
- Episodic, semantic and procedural memoryEpisodic memory records specific events, semantic memory stores general facts, and procedural memory holds how to do things; AI systems borrow all three.
- Knowledge graph memory vs vector memoryKnowledge graph memory stores facts as linked entities and relationships; vector memory stores text chunks as embeddings and finds them by similarity.
- ChunkingChunking is splitting a document into smaller passages before embedding, so each piece can be indexed, searched and returned on its own.
See the terms in practice.
Central Brain keeps project memory in files on storage you choose and serves selected context to supported AI tools over MCP. Licensing uses Keygen. $12 a month per license.