RAG is an AI architecture that enhances Large Language Model outputs by retrieving relevant information from external knowledge bases before generating responses. Instead of relying solely on training data, RAG systems search company documents, databases, or websites to ground their answers in current, specific information. This approach dramatically reduces hallucinations and makes AI practical for enterprise use cases like customer support, knowledge management, and internal search.
DE: RAG ist eine KI-Architektur, die LLM-Ausgaben durch Abruf relevanter Informationen aus externen Wissensbasen verbessert. Dieser Ansatz reduziert Halluzinationen und macht KI für Unternehmensanwendungen praxistauglich.