AI & Knowledge Systems Glossary
Essential terminology for AI agents, knowledge graphs, RAG systems, and enterprise memory platforms. From vector databases to temporal memory, entity linking to business outcomes—comprehensive definitions for developers and technical leaders.
Core Concepts
Foundational concepts for AI agents and knowledge systems
Agent Memory
Persistent, structured memory that allows AI agents to recall prior interactions, decisions, and context over time.
Agent Memory Platform
A system that provides ingestion, storage, retrieval, and reasoning layers so agents can maintain durable, time-aware context across workflows.
Agent Memory Framework
A conceptual and architectural model for implementing memory into autonomous agents, including state, timeline, and relationship structures.
Operating Memory
The shared, continuously updated context that reflects what a team or system is doing, what changed, and what needs to happen next.
Stateful Agent
An autonomous agent that retains memory of previous steps, outcomes, and reasoning, enabling cumulative learning and improvement.
Memory Types
Understanding different memory architectures and their use cases
Semantic Memory
Meaning-based memory that stores relationships and concepts, enabling reasoning and synthesis.
Temporal Memory
Memory that tracks how state changes over time, enabling agents to understand sequences, causality, ownership, and status changes.
Session Memory
Short-lived context that lasts during a conversation or run, but disappears afterward — unlike persistent agent memory.
Ephemeral Context
Temporary reasoning state inside a prompt — descriptive but not durable; resets every interaction.
Technical Infrastructure
Core technologies powering AI systems and knowledge platforms
Knowledge Graph
A structured representation of entities and relationships used to power reasoning, personalization, and contextual awareness.
Context Window
The limited amount of information an LLM can consider at once; why pure prompt-based memory resets and does not persist state.
Vector Database
A system that stores embeddings for similarity search — useful for retrieval, but insufficient for time-aware structured memory.
Context Engine
A service that assembles the most relevant structured memory for a task or agent state, dynamically and with recency awareness.
Memory Index
The lookup layer that allows agents to retrieve relevant prior context without reprocessing entire histories.
Operations & Processes
Essential processes for building and maintaining knowledge systems
Entity Linking
The process of recognizing when multiple references point to the same person, project, document, or concept in memory.
Event Timeline
A chronological record of work, interactions, decisions, and changes that agents use to contextualize and anticipate next steps.
Fact Extraction
Identifying and storing key claims, decisions, tasks, and structured data from text, audio, and collaborative workflows.
Semantic Retrieval
Fetching context based on meaning and relationships, not just keyword matches or vector similarity alone.
Comparisons & Distinctions
Understanding key differences between AI approaches and architectures
Business Outcomes
Real-world applications and business value of AI knowledge systems
Continuity of Work
The ability for work, ownership, and decisions to persist across contexts, documents, and time.
Status Brief
A synthesized summary of what happened, what changed, and what needs to happen next.
Operating Review
A recurring review process informed by structured memory rather than manual reporting or recall.
Knowledge Fabric
An interconnected network of organizational memory that updates continuously as work happens.
Progress Tracking
Monitoring how work evolves over time, including dependencies, ownership, and status changes.