Skip to content

Retrieval-Augmented Generation

Definition

Retrieval-augmented generation, or RAG, is a method where an AI system retrieves relevant information before generating an answer or taking action.

RAG grounds AI outputs in a company's documents, knowledge bases, policies, product information, customer history, or other trusted sources. Instead of relying only on model memory, the system pulls current context into the task.

For agent systems, RAG is often the difference between useful and dangerous. An agent handling support, sales, onboarding, or internal questions needs access to the right information at the right moment. Without retrieval, it guesses. With retrieval, it can operate from company-specific truth.

Related terms

AI Memory

AI memory is the ability of an AI system to retain useful context across interactions, tasks, users, or workflows.

AI Tool Use

AI tool use is the capability of an AI system to select and use external tools to complete a task.

Long-Term Memory for Agents

Long-term memory for agents refers to persistent memory that allows agents to retain information across sessions and over time.

Model Context Protocol

Model Context Protocol, often called MCP, is an emerging standard for connecting AI systems to tools, data sources, and external context.