Glossary
The definitive glossary of AI agent memory terms and concepts. From episodic memory to vector databases, understand the building blocks of intelligent agents.
A(6)
Understand the Agent Loop, the core execution cycle of AI agents: perceive, reason, decide, execute, and observe.
Agent memory enables AI systems to retain and recall information across conversations. Learn why memory is essential for intelligent AI agents.
Learn about Agent State, the complete snapshot of an AI agent's current situation including context, memory, and progress.
Learn about Agentic RAG, a retrieval-augmented generation pattern where AI agents dynamically decide when, what, and how to retrieve information.
Understand attention mechanisms, neural network components that allow models to focus on relevant parts of input sequences.
Explore AutoGPT, an open-source autonomous AI agent framework that chains together LLM calls to accomplish complex tasks.
B(3)
Learn about BabyAGI, a lightweight AI agent framework that maintains and executes a dynamic task list to achieve goals.
Understand BM25, a ranking function used in information retrieval based on term frequency and document length normalization.
Learn about Buffer Memory, a memory implementation that stores a fixed number of recent interactions in agent systems.
C(15)
Discover Chain of Thought prompting, a technique that guides LLMs to show step-by-step reasoning for better problem-solving.
Explore Chroma, an open-source embedding database designed for AI applications and vector similarity search.
Understand chunk overlap, the practice of sharing tokens between adjacent chunks to preserve context in document segmentation.
Learn about chunk size, the target length of document segments in text splitting for embedding and retrieval systems.
Learn about chunking, the process of dividing large documents into smaller segments for efficient storage and retrieval in AI systems.
Explore combined memory architectures that merge multiple memory types to provide both immediate and historical context for AI agents.
Understand context length, the maximum number of tokens a language model can process in a single request.
Learn about context windows, the span of tokens a language model can attend to when processing and generating text.
Explore Conversation Buffer Memory, a LangChain class that stores complete conversation history for AI agents.
Learn about Conversation Memory, the system for tracking dialogue history in conversational AI agents.
Discover Conversation Summary Memory, a memory type that maintains a running summary of conversation history for efficient context management.
Learn about Conversation Token Buffer Memory, a buffer implementation that tracks token usage to stay within model context limits.
Explore Corrective RAG, an advanced RAG pattern that evaluates retrieval quality and takes corrective actions to improve results.
Understand cosine similarity, a metric for measuring semantic similarity between vector embeddings in AI systems.
Discover CrewAI, a framework for orchestrating multiple AI agents that collaborate to accomplish complex tasks.
D(6)
Learn about data ingestion, the process of loading and preparing documents for storage in AI agent memory systems.
Understand deduplication, the process of identifying and removing duplicate entries from knowledge bases and memory systems.
Explore Dense Passage Retrieval, a technique using trained bi-encoder models to retrieve passages based on semantic similarity.
Learn about dense retrieval, a technique that uses embeddings to find semantically similar content in vector space.
Understand document embeddings, vector representations that capture the semantic meaning of entire documents.
Learn about document loaders, components that extract and structure text content from various file formats for AI systems.
E(7)
Embeddings convert text into numerical vectors for semantic search. Learn how AI agents use embeddings for memory retrieval.
Explore embedding models, neural networks that transform text into dense vector representations for semantic processing.
Learn about entity extraction, the process of identifying and extracting structured information about entities from unstructured text.
Discover entity memory, a memory type that tracks structured information about specific entities mentioned in agent interactions.
Understand entity stores, databases designed for storing and querying structured information about entities and their attributes.
Episodic memory stores specific events and experiences. Learn how AI agents use episodic memory to recall past interactions and context.
Learn about Euclidean distance, a metric for measuring the straight-line distance between vectors in embedding space.
F(4)
Explore FAISS, Facebook AI's high-performance library for efficient similarity search and clustering of dense vectors.
Learn about few-shot learning, a technique where models learn to perform tasks from a small number of examples provided in the prompt.
Understand fine-tuning, the process of training pre-trained models on specific datasets to specialize their behavior.
Discover function calling, an LLM capability that enables models to request structured tool use and API invocations.
G(2)
H(5)
Understand hallucinations, instances where language models generate incorrect or fabricated information presented as fact.
Explore Haystack, an open-source framework for building production-ready RAG and search applications.
Discover hybrid memory architectures that combine multiple memory storage and retrieval methods for agent systems.
Learn about Hybrid RAG, a retrieval approach combining dense and sparse retrieval for improved accuracy.
Explore hybrid search, a technique combining vector-based semantic search with keyword-based retrieval methods.
I(4)
Understand in-context learning, the ability of language models to adapt to new tasks from examples or instructions in prompts.
Learn about indexing, the process of organizing data structures to enable efficient retrieval and similarity search.
Discover instruction tuning, a training technique that teaches models to follow natural language instructions effectively.
Understand inverted indexes, data structures that map terms to documents for efficient keyword-based search.
K(4)
Learn about knowledge bases, organized repositories of information that AI agents can query and retrieve from.
Explore knowledge graphs, structured representations of information as interconnected entities and relationships.
Learn about knowledge triples, the fundamental units of knowledge graphs representing facts as subject-predicate-object structures.
Discover triple stores, specialized databases optimized for storing and querying knowledge triples and graph-structured data.
L(3)
Explore LangChain, a popular framework for building LLM-powered applications including agents and RAG systems.
Learn about LlamaIndex, a data framework for connecting custom data sources to large language models.
Long-term memory enables AI agents to retain information across sessions. Learn about persistent memory architectures for intelligent agents.
M(16)
Learn about memory in AI systems, the capability for agents to retain and utilize information from previous interactions.
Learn about memory banks, storage systems for organizing and managing multiple memories in AI agent architectures.
Memory consolidation transforms short-term memories into efficient long-term storage. Learn how AI agents compress and optimize their memory.
Understand memory decay, the reduction in memory importance or accessibility over time in agent systems.
Learn about memory encoding, the process of converting information into storable formats for agent memory systems.
Explore memory hierarchy, the organization of memory systems into levels based on access speed and capacity.
Understand memory indexes, data structures that enable efficient lookup and retrieval of stored memories.
A memory layer provides persistent memory infrastructure for AI applications. Learn about memory layer architecture and design patterns.
Learn about LangChain memory modules, components that manage state and context across agent interactions.
Understand memory persistence, the capability for agent memories to survive beyond individual sessions.
Learn about memory retrieval, the process of finding and accessing relevant stored information in agent memory systems.
Explore memory streams, chronological logs of observations and experiences used in agent architectures like Generative Agents.
Understand metadata filtering, the technique of narrowing search results based on document attributes in retrieval systems.
Explore Milvus, an open-source vector database designed for scalable similarity search on massive embedding datasets.
Learn about multi-head attention, a mechanism that runs multiple attention operations in parallel for richer representations.
Discover multi-hop retrieval, a technique that chains multiple retrieval steps to answer complex questions requiring information synthesis.
N(3)
Learn about Named Entity Recognition, the task of identifying and classifying named entities in text.
Explore Neo4j, a leading graph database platform for storing and querying knowledge graphs and graph-structured data.
Understand normalization, the process of standardizing text or vectors to improve consistency and retrieval quality.
O(1)
P(9)
Discover pgvector, a PostgreSQL extension that adds vector similarity search capabilities to relational databases.
Learn about Pinecone, a fully managed vector database service for building semantic search and AI applications.
Understand planning in AI agents, the process of decomposing goals into actionable steps and strategies.
Learn about positional encoding, the mechanism that injects sequence order information into transformer models.
Understand postprocessing, the operations performed after retrieval to refine and optimize results before using them.
Learn about preprocessing, the operations that clean and prepare text before embedding or storage in retrieval systems.
Explore procedural memory, the memory of skills, procedures, and how to perform tasks in AI agent systems.
Learn about prompt engineering, the practice of designing effective prompts to elicit desired behaviors from language models.
Understand property graphs, graph structures where nodes and edges can have multiple typed properties and labels.
Q(3)
Discover Qdrant, a vector database with advanced filtering and hybrid search capabilities for AI applications.
Learn about query expansion, the technique of augmenting queries with additional terms or variations to improve retrieval.
Understand query rewriting, the technique of transforming user queries into more effective search formulations.
R(12)
RAG combines retrieval with generation for grounded AI responses. Learn how agents use RAG for memory-enhanced conversations.
Explore RAG Fusion, a technique that generates multiple query variations and combines their results for improved retrieval.
Learn about RDF (Resource Description Framework), a standard for representing information as subject-predicate-object triples.
Discover re-ranking, a two-stage retrieval approach that refines initial results using more sophisticated scoring models.
Learn about ReAct (Reasoning and Acting), a paradigm that interleaves reasoning and action execution in AI agents.
Understand read-only memory, static knowledge sources that agents can retrieve from but not modify.
Learn about reflection in AI agents, the capability to analyze and learn from past experiences and actions.
Understand relation extraction, the process of identifying and classifying relationships between entities in text.
Learn about relevance scores, numerical measures of how well retrieved results match query intent in retrieval systems.
Understand retrieval accuracy, the measure of how well a retrieval system finds relevant information and avoids irrelevant results.
Learn about retrieval pipelines, the sequence of processing steps from query to final retrieved results in RAG systems.
Discover RLHF (Reinforcement Learning from Human Feedback), a technique for aligning AI models with human preferences and values.
S(17)
Learn about scratchpads, temporary working memory spaces where agents can store and manipulate intermediate results.
Understand self-attention, a mechanism where sequences attend to themselves to capture internal dependencies and relationships.
Learn about self-critique, the capability for agents to evaluate and improve their own outputs through iterative refinement.
Discover Self-RAG, a technique where models dynamically decide when to retrieve and self-evaluate their outputs.
Explore Semantic Kernel, Microsoft's SDK for integrating LLMs with conventional programming for AI applications.
Semantic memory stores facts, concepts, and general knowledge. Learn how AI agents use semantic memory for reasoning and knowledge retrieval.
Learn about semantic search, a retrieval technique that finds results based on meaning rather than exact keyword matching.
Understand sentence embeddings, vector representations that capture the semantic meaning of complete sentences.
Learn about shared memory, memory systems accessible by multiple agents for coordination and collaboration.
Understand short-term memory in AI agents, the temporary storage of recent information and immediate context.
Learn about similarity search, the process of finding items most similar to a query in vector space.
Discover SPARQL, a query language for retrieving and manipulating data stored in RDF format.
Understand sparse retrieval, keyword-based search using term matching and inverted indexes for efficient retrieval.
Learn about stateful agents, AI systems that maintain and build upon state across multiple interactions.
Understand stateless agents, AI systems that treat each interaction independently without maintaining state.
Learn about summary memory, a memory type that compresses information into condensed summaries for efficient storage.
Learn about system prompts, the initial instructions that define an AI agent's behavior, role, and capabilities.
T(8)
Understand taxonomies, hierarchical classification systems that organize concepts into parent-child relationships.
Learn about text splitting, the process of dividing documents into manageable segments for processing and storage.
Understand TF-IDF, a statistical measure for evaluating term importance in documents for information retrieval.
Learn about tokens, the basic units of text that language models process and generate.
Understand tokenization, the process of converting text into sequences of tokens for language model processing.
Learn about tool use, the capability for AI agents to invoke external functions, APIs, and services to extend their capabilities.
Explore transformers, the neural network architecture that revolutionized NLP and powers modern language models.
Learn about triple stores, databases optimized for storing and querying knowledge represented as subject-predicate-object triples.
V(2)
Vector databases store embeddings for semantic similarity search. Learn how AI agents use vector stores for memory retrieval.
Understand Vector Store Memory, a LangChain memory type using semantic search to retrieve relevant past interactions.