Definition
Also written: RAG
Retrieval-augmented generation is an architecture where a language model answers using documents fetched at query time rather than relying only on training data. The system splits sources into chunks, embeds and ranks them, places the best matches into the model’s context, and generates an answer from those.
Understanding this explains most credible advice about writing for AI discovery. Because documents are chunked before retrieval, the unit that gets embedded, ranked and quoted is the passage — which is why self-contained paragraphs matter mechanically, not just stylistically.
Research on RAG pipelines also found that stripping HTML to plain text loses structural signal that improves retrieval, arguing for real headings, tables and lists over visually-identical div markup.
The usual error
Where this goes wrong in practice
Assuming a model "remembers" a site. In a retrieval system, content is fetched per query — which means it has to be reachable and readable at that moment.
Related terms
Generative engine optimization (GEO)
Generative engine optimization is the practice of improving how a source is used by AI systems that synthesise answers rather than list links. The term comes from a 2023 academic paper (arXiv:2311.09735, later accepted at KDD 2024) that measured how often a document’s content was reflected in a generated answer.
DefinitionAI Overviews
AI Overviews are Google-generated summaries shown above traditional results for some queries, assembled from pages in the ordinary search index and shown with links to the sources used. Eligibility requires that a page is already indexable and snippet-eligible in normal Search.
Definitionllms.txt
llms.txt is a proposed convention: a Markdown file at a site’s root listing its important pages for AI systems. Adoption has grown quickly, but current measurement shows AI crawlers largely ignore it and fetch HTML directly instead.
DefinitionAgentic workflow
An agentic workflow is an automation where a language model completes several connected steps toward a goal — researching, classifying, drafting, routing, reporting — deciding the order within defined rules, rather than executing a fixed script. Human approval points are part of the design, not an afterthought.
DefinitionAnswer engine optimization (AEO)
Answer engine optimization is structuring content so a direct answer can be lifted cleanly into a featured snippet, a People Also Ask entry, a voice response or an AI summary. It predates the current AI wave — the original concern was voice assistants — and overlaps heavily with both SEO and GEO.
DefinitionLLMO, AISO and GSO
LLMO, AISO and GSO are vendor terms for largely the same work as GEO and AEO: making content that AI systems can read, trust and reference accurately. No academic or platform-official definition separates them, and industry sources contradict each other on where the boundaries fall.
DefinitionPrompt injection
Prompt injection is an attack where instructions hidden in content an AI system reads — a web page, an email, a document, a form submission — are treated by the model as commands rather than data. It is the main security concern for any agent that ingests untrusted input.
Definition