→ The dictionary
Glossary
Every term, defined once, in citeable shape. Built so this page itself ranks for the queries it explains.
- AIO (AI Optimization)
- The discipline of making content discoverable, citeable, and faithfully represented by AI answer engines.
- AI Optimization (AIO) is the umbrella discipline for engineering content, schema, and distribution so that AI Overviews, ChatGPT, Perplexity, Claude, Gemini, and Apple Intelligence cite you accurately. It encompasses GEO, AEO, ASO, and emerging surfaces like agent-to-agent commerce.
- See also: geo, aeo, aso
- GEO (Generative Engine Optimization)
- SEO for generative engines: optimizing for citation and inclusion in AI-generated answers.
- GEO is the practice of optimizing content to be cited by generative answer engines (Perplexity, ChatGPT Search, Google AI Overviews). Unlike classical SEO, success is measured in citation share, not click-throughs.
- See also: aio, aeo
- AEO (Answer Engine Optimization)
- Optimizing specifically to be the answer — not just to be cited.
- AEO focuses on owning the direct answer surface. Where GEO accepts being one of many citations, AEO targets the verbatim answer text the engine produces. Heavy emphasis on atomic facts, FAQPage schema, and HowTo structure.
- See also: geo, aio
- ASO (AI Summary Optimization)
- Optimizing for the 1–2 sentence summary surfaces of Apple Intelligence and Gmail Gemini.
- ASO is the newest layer: writing email subjects, preview text, and body copy so that the AI-generated 1–2 sentence summary (now shown before the email itself in Apple Mail and Gmail) accurately and persuasively represents the message. Take One Glance was built for this surface.
- See also: aio
- AI Overview
- Google's generative answer block at the top of a search results page.
- AI Overviews (formerly SGE) are AI-generated summaries Google shows above traditional results for an estimated 30–50% of queries. They cite sources inline and are the dominant new SEO surface.
- See also: ai-mode
- AI Mode
- Google's dedicated conversational search experience, built on Gemini.
- AI Mode is Google's standalone conversational search surface, separate from the AI Overview block on traditional SERPs. It supports multi-turn dialogue and deeper synthesis.
- llms.txt
- A root-level file telling LLM crawlers which pages on your site to read first.
- Proposed convention: /llms.txt is to LLM crawlers what robots.txt is to search crawlers. A markdown file with a site summary and curated links to your most authoritative pages.
- Atomic Facts
- One self-contained claim per sentence, optimized for retrieval-chunk extraction.
- RAG retrievers chunk content into short passages. Atomic-fact writing ensures each chunk contains a complete, citeable claim — dramatically increasing odds of verbatim inclusion in an AI answer.
- Citeable Quote
- A named, attributed quote in semantic markup that LLMs prefer to surface.
- LLMs weight quotable, attributable strings higher because they reduce hallucination risk. <blockquote cite='...'> with Person schema attribution is the canonical form.
- JSON-LD
- The structured data format Google and every LLM crawler prefers.
- JSON-LD is a JSON-based serialization of Linked Data. Embedded in <script type='application/ld+json'> blocks, it gives crawlers an unambiguous machine-readable description of the page's entities.
- RAG (Retrieval-Augmented Generation)
- The architecture behind every AI answer engine: retrieve relevant docs, then generate.
- RAG systems first retrieve candidate passages from an index, then condition generation on those passages. Understanding RAG explains why atomic facts, named entities, and schema all matter — they shape what gets retrieved.
- Answer Box Autopsy
- A teardown of why a specific page won (or lost) a specific AI answer.
- The core unit of analysis on AI Hacked. Document the prompt, surface, winner, the trick they used, and how to reproduce or counter it.
- Agent as Audience
- Designing content for an AI agent reader, not a human reader.
- The framing shift behind agent-first press releases, llms.txt files, and machine-readable boilerplates. The journalist isn't the first reader anymore. The agent is.