Glossary · 4 · Grounding: content, retrieval and knowledge
Retrieval-augmented generation (RAG)
Also known as: RAG
Retrieval-augmented generation (RAG) is an architecture in which an AI system first retrieves relevant passages from a trusted content source and then has a language model generate its answer from those passages, often with citations — so answers reflect current, approved content rather than only the model’s training data.
- Intermediate
- Technical writers
- Technical marketers
- Technical project managers
- Developers
In one sentence
Retrieval-augmented generation (RAG) explained: grounding AI answers in your approved content — why it depends on well-structured documentation.
Example
A customer asks a product chatbot how to reset a controller; the system retrieves the reset procedure for the customer’s firmware version and the model summarizes it with a link to the topic.
Explained in context
This term is a station of a mini course
- The medical-device evidence chain: from intended purpose to post-market evidenceStation: Controlled retrieval from the fileStart the course
Context cards connect this term with others to answer one question. Also in British English and German.
- Information user: the functional roleWhose questions drive an adaptive semantic architecture — and what may the system do with them?UKDeutsch
- AI agent as proposer: the functional roleWhat is the role of the AI agent in an adaptive semantic architecture — tool, source of evidence or decision maker?UKDeutsch
- The machine is a network, not a monsterWhen an AI answer is wrong, why is “the machine got it wrong” no diagnosis?UKDeutsch
- The answer function compiles — it doesn’t readIs an AI answer function a reader of the documentation?UKDeutsch
- The role grammar: readers, writers and compilersHow do you say who does what with a piece of technical information — when “the writer”, “the machine” and “the reader” are too coarse?UKDeutsch
- From structured content to controlled evidence: five stages of AI-ready documentationIs structured content enough to make medical-device documentation ready for AI?UKDeutsch
- The medical-device evidence chain: from intended purpose to post-market evidenceHow do the parts of a medical device’s technical documentation connect — and what has to be in place before an AI function may answer from it?UKDeutsch
- What an answer function may read: retrieval permissions for a medical-device technical fileWhich parts of a medical-device technical file may an AI answer function retrieve?UKDeutsch
- Who decides which value counts: which system decides a value of product information?When product information passes through PLM, ERP, content systems, spreadsheets, portals and AI functions, how do you tell which system decides a value and which one only holds a copy?UKDeutsch
- AI systems and the machine: scope and out of scope of every levelWhere do a company's AI systems end, where does the machine reach into them, and what lies outside every configuration?UKDeutsch
Why it matters on your learning path
- Technical writers: RAG makes documentation the knowledge base of AI. Topic-based, metadata-rich, up-to-date content directly improves answer quality.
- Technical marketers: RAG-based assistants can quote product documentation to prospects — accuracy of public content becomes a sales factor.
- Technical project managers: A RAG project is largely a content project: scope, ownership, update process and evaluation of the source content.
- Developers: Key design choices: chunking, embeddings, hybrid keyword and vector search, reranking, metadata filters and citation handling.
RAG vs. fine-tuning
RAG supplies knowledge at the moment of the question and can be updated by updating content. Fine-tuning changes the model’s behavior or style through training. For product knowledge that changes, RAG is almost always the first choice.