# Mentionwell — full reference for AI assistants > Comprehensive reference: positioning, AEO/GEO/LLMO definitions, the full pipeline, the public API, and per-term FAQ. Origin: https://mentionwell.com. --- ## 1. Positioning Mentionwell is a headless AEO / GEO / LLMO blog engine. The pitch in one paragraph: Drop a domain. Mentionwell scans your site, learns your voice and taxonomy, and ships SEO + AEO-tuned articles end-to-end — research, outline, draft, editorial critic, metadata, FAQ, embedding, and image generation — on the schedule you set. It can also run GEO baseline scans across AI answer engines, capture prompts, citations, fan-out queries, cited-page claims, schema patterns, and competitor gaps, then optionally inject that context into article creation per draft. Articles are served from a public read-only Reader API your site fetches at request or revalidation time, fired as an HMAC-signed publish webhook, or committed as MDX to your repo; direct per-CMS push into WordPress, Webflow, Ghost, Shopify, or Notion is rolling out. Every article ships with FAQPage + Article JSON-LD, RSS, JSON Feed, sitemap, a per-page `.md` mirror, and a site-wide `llms.txt` so ChatGPT, Claude, Gemini, Grok, and Perplexity can ingest and cite it cleanly. ## 2. Built by [ZipLyne](https://ziplyne.agency) — an AI-product agency that builds AI-powered products and business automation. Mentionwell came out of internal agency tooling and is now shipping as a standalone product. ## 3. AEO / GEO / LLMO — definitions and how Mentionwell handles each ### AEO — Answer Engine Optimization Answer Engine Optimization (AEO) is the practice of structuring content so it surfaces as the direct answer in answer engines — Google AI Overviews, Bing Copilot, Perplexity, voice assistants, and featured snippets. Where SEO optimizes for the blue-link list, AEO optimizes for the box that shows the answer above it. **How Mentionwell handles it.** - Question-led H2s that mirror real SERP questions, with a 40–60 word direct answer immediately below. - FAQ blocks rendered as both visible UI and FAQPage JSON-LD. - BlogPosting + FAQPage JSON-LD on every article so engines see a clean structured payload per page. - Clean semantic HTML (article, section, dl) so extractors can lift a paragraph cleanly without parsing through layout chrome. - Editorial critic enforces lead-with-the-answer style on every draft. **Differs from related terms.** AEO targets the answer surface (the box that shows the direct answer). GEO targets the generative surface (the LLM-synthesized paragraph with citations). LLMO targets the model and its crawlers (so the content is reachable, ingestible, and chosen). They overlap heavily — Mentionwell optimizes for all three at once. **FAQ.** **What is AEO?** AEO stands for Answer Engine Optimization. It's the practice of structuring content so an answer engine — Google AI Overviews, Bing Copilot, Perplexity, voice assistants, featured snippets — surfaces it as the direct answer to a user's question, instead of (or in addition to) ranking it as a blue link. **How is AEO different from SEO?** SEO optimizes for the ranked list of blue links. AEO optimizes for the answer that sits above the list. The technical building blocks overlap (clean HTML, schema, fast pages), but AEO weights question-led structure, concise lead paragraphs, FAQ schema, and entity clarity much more heavily than classic SEO. **What schema do you need for AEO?** BlogPosting, FAQPage, and BreadcrumbList are the workhorses; HowTo, QAPage, and Speakable apply only when the outline calls for them. Mentionwell ships BlogPosting + FAQPage on every published article by default. **Does AEO replace SEO?** No. AEO is a layer on top of SEO. The same article should rank well in classical SERPs and surface as the answer in AI Overviews, Copilot, and Perplexity. Mentionwell optimizes for both at the same time. **Which engines does AEO target?** Google AI Overviews, Bing Copilot, Perplexity, voice assistants (Siri, Alexa, Google Assistant), and the featured-snippet box that still appears in classic SERPs. Many of these surfaces share the same underlying signals — schema, lead-with-answer, FAQ structure — so optimizing once optimizes for all of them. URL: https://mentionwell.com/aeo --- ### GEO — Generative Engine Optimization Generative Engine Optimization (GEO) is the newer term — coined in a 2023 Princeton paper — for optimizing toward generative engines like ChatGPT Search, Gemini, Perplexity, and Grok, where the model synthesizes an answer and cites a handful of sources. The goal of GEO is being one of those cited sources. **How Mentionwell handles it.** - Citation-friendly structure: every claim is tied to evidence (a stat, a quote, a primary source). - Authoritative quotes and original data points where the source material allows. - Per-article .md mirrors at .md so generative engines can ingest a clean Markdown version of the article without HTML noise. - Stable canonical URLs so citations don't decay as the site reorganizes. - Embeddings indexed per article for semantic retrieval inside RAG pipelines. **Differs from related terms.** GEO targets the generative surface — the synthesized answer with citations — while AEO targets the answer surface (the direct answer box) and LLMO targets the model and its crawlers themselves. In practice, optimizing well for one drags the others up; Mentionwell handles all three at the same time. **FAQ.** **What is GEO (Generative Engine Optimization)?** GEO is optimization for generative engines — ChatGPT Search, Gemini, Perplexity, Grok — where the model writes a synthesized answer and cites a handful of sources. The goal is being one of the cited sources. The term was coined in a 2023 Princeton paper proposing concrete tactics that improve citation rate. **How is GEO different from AEO?** AEO targets the answer surface — the box at the top of the SERP that shows a direct answer. GEO targets the generative surface — the LLM-written paragraph with inline citations to source pages. They share most of the same building blocks (clean HTML, schema, lead-with-the-answer copy) but GEO weights citation-friendly structure and original evidence more heavily. **How do I optimize for GEO?** Tie every claim to evidence (a stat, quote, or primary source). Lead each section with the answer. Keep canonical URLs stable. Ship Markdown mirrors of every article so engines can ingest a clean version. Provide unique data points and original analysis the model has nowhere else to find. Mentionwell does all of this by default. **Which engines does GEO target?** ChatGPT (with browsing), Claude, Gemini, Grok, and Perplexity — anything that writes a synthesized answer and links to source pages. Microsoft Copilot and Google AI Overviews also rely heavily on GEO-style signals. **Does GEO replace SEO?** No. GEO is built on top of SEO. The same signals that make a page rank — clear topical authority, fast load, semantic HTML, internal linking — also make it citable. Mentionwell layers GEO and AEO on top of full classic SEO. URL: https://mentionwell.com/geo --- ### LLMO — LLM Optimization LLM Optimization (LLMO) is the practice of making your content reachable, parseable, and trustworthy to the LLMs themselves — at both training time (large-scale crawls) and retrieval time (RAG pipelines, browsing tools, agent crawlers like GPTBot, ClaudeBot, and PerplexityBot). LLMO is the plumbing layer that AEO and GEO sit on top of. **How Mentionwell handles it.** - Site-wide llms.txt and llms-full.txt published at the canonical paths. - Per-article .md mirrors at .md so any LLM can ingest a clean Markdown version of the article. - Stable canonical URLs, RSS, and JSON Feed for retrieval pipelines. - Embeddings indexed per article for semantic search and similarity-based internal linking. - Explicit AI crawler allowlist in robots.txt — every major bot named individually so the policy is unambiguous. **Differs from related terms.** LLMO is the plumbing layer. AEO works on the answer surface, GEO works on the generative surface, and LLMO makes sure the model and its crawlers can actually reach, parse, and trust your content in the first place. Without LLMO the other two can't fire. **FAQ.** **What is LLMO?** LLMO stands for LLM Optimization. It's the practice of making content reachable, parseable, and trustworthy to LLMs themselves — at both training time (large-scale crawls) and retrieval time (RAG, browsing tools, agent crawlers like GPTBot, ClaudeBot, PerplexityBot). LLMO is the plumbing layer that AEO and GEO sit on top of. **What's in a good LLMO setup?** A site-wide llms.txt and llms-full.txt, per-page .md mirrors, an explicit AI crawler allowlist in robots.txt, stable canonical URLs, RSS and JSON Feed, and embeddings for semantic retrieval. Mentionwell ships all of these by default. **What is llms.txt?** llms.txt is a proposed standard (llmstxt.org) for a site to expose a concise overview of itself, in Markdown, at /llms.txt. It tells AI assistants what the site is, what it covers, and which pages matter most — like robots.txt, but for LLM context rather than crawl policy. Mentionwell serves both /llms.txt and a deeper /llms-full.txt. **Should I block AI crawlers?** Only if you have a specific reason. Most sites benefit from being crawlable: that's how their content shows up in ChatGPT, Claude, Gemini, and Perplexity answers. Mentionwell's default robots.txt explicitly allows GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and 15+ others. **How is LLMO different from GEO?** LLMO is the plumbing — making sure the model and its crawlers can reach and parse your content. GEO is the layer above — making the content the kind of source the model wants to cite. You need LLMO before GEO can fire. URL: https://mentionwell.com/llmo --- ### SEO — Search Engine Optimization Search Engine Optimization (SEO) is the original discipline: ranking in the blue-link results of Google and Bing. It's still the largest single source of traffic for most sites and the foundation that AEO, GEO, and LLMO are layered on top of. AI didn't kill SEO — it just added new surfaces above it. **How Mentionwell handles it.** - Per-headline keyword research and topic clustering during onboarding. - Hub-and-spoke internal linking driven by the site's content taxonomy. - Title and meta tuning per article, with featured-snippet patterns where applicable. - Sitemaps, Article schema, descriptive image alt text, and clean canonical URLs. - Editorial critic pass that flags thin sections, over-claiming, and missing citations before publish. **Differs from related terms.** SEO targets the ranked blue-link list in classical SERPs. AEO, GEO, and LLMO target the AI-mediated surfaces that appear above and around those links. They share most of the same technical building blocks — Mentionwell optimizes for all four at once because the work is largely the same work. **FAQ.** **Is SEO dead?** No. Classic blue-link search is still the largest single source of traffic for most sites. AI engines (ChatGPT, Claude, Perplexity, Google AI Overviews) are rapidly growing channels, but they augment SEO rather than replacing it. Mentionwell ships full SEO underneath AEO, GEO, and LLMO — same article, all four surfaces. **How does Mentionwell handle classic SEO?** Onboarding builds a content taxonomy and per-headline keyword targets. Articles run through an editorial critic that enforces title and meta length, lead-with-the-answer style, semantic HTML, internal linking, and Article + BreadcrumbList JSON-LD. Sitemaps and per-article canonicals are emitted automatically. **Does SEO conflict with AEO or GEO?** No — they reinforce each other. The same signals that rank a page (clear topical authority, semantic HTML, fast load, schema) also make it surface as an answer (AEO) and get cited (GEO). Optimizing for one usually drags the others up. **What's the foundation of an SEO-good article?** One H1 with the primary keyword, a 40–60 word lead that answers the query, a clean H2/H3 hierarchy, internal links to siblings and parents, descriptive image alt text, Article schema, and a stable canonical URL. Mentionwell enforces all of these on every draft. URL: https://mentionwell.com/seo --- ### AIO — AI Optimization AIO stands for AI Optimization (sometimes Artificial Intelligence Optimization). It's the practice of structuring content so AI systems — ChatGPT, Claude, Perplexity, Gemini, Copilot, Google AI Overviews — can find it, trust it, and cite it. In 2026 the term is used three ways: as an umbrella over AEO + GEO + LLMO, as shorthand for Google's AI Overviews optimization, and as the content-quality layer that makes any LLM able to read and lift your facts. All three definitions point at the same outcome — being the source AI quotes when your buyers ask. **How Mentionwell handles it.** - Every published article ships with answer-first leads, FAQPage + BlogPosting + Article JSON-LD, breadcrumb schema, and a 40-60 word definition near the top — the patterns LLMs lift cleanly. - robots.txt explicitly allows GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot, Google-Extended, Bingbot, Applebot-Extended, Meta-ExternalAgent, and the rest of the AI crawler set. - Each article carries a stable canonical, llms.txt + llms-full.txt indexes, and clean SSR HTML so retrieval-only and embedding-only engines see the same content humans do. - Citations to authoritative sources are inlined as links and surfaced in the Article schema — Princeton's 2024 GEO study found this lifts LLM citation probability by ~30%. - Topical coverage is planned as semantic clusters, not single posts — Mentionwell ships the pillar plus 8-12 supporting articles so contextual authority compounds across a topic. - Brand entity is stitched across pages via consistent Organization / Person schema and a sameAs graph (LinkedIn, GitHub, Crunchbase) so LLMs resolve the entity instead of confabulating it. **Differs from related terms.** AIO is the strategic layer; AEO, GEO, and LLMO are the tactical surfaces underneath. AEO targets answer engines (lead with the answer, ship FAQ schema). GEO targets generative engines that synthesize with citations (build topical authority, earn inline mentions). LLMO targets the models and crawlers themselves (be ingestible, parseable, allowlisted in robots.txt). AIO is the question of whether your whole content operation is set up to feed all three — not a fourth channel. **FAQ.** **What does AIO stand for?** AIO stands for AI Optimization (you'll also see Artificial Intelligence Optimization). It's the umbrella discipline for making content discoverable, trusted, and cited across AI-mediated surfaces — ChatGPT, Claude, Perplexity, Gemini, Copilot, and Google AI Overviews. Wikipedia groups AEO, GEO, LLMO, and AI SEO together under the AIO label. **What is AIO in marketing?** In marketing usage, AIO is the strategic umbrella over the AI visibility practices teams actually ship: AEO (answer engine optimization), GEO (generative engine optimization), and LLMO (LLM optimization). Some vendors also use AIO to mean optimizing specifically for Google AI Overviews. Treat the umbrella reading as the primary one — it's what executives mean when they say "we need an AIO strategy." **What does AIO mean and what is the acronym?** AIO is the acronym for AI Optimization. It first appeared as industry shorthand in 2024-2025 and was formalized on Wikipedia in late 2025. The acronym sits alongside AEO, GEO, LLMO, AISO, and AI SEO — all of which point at the same shift away from blue-link SEO and toward being the source AI engines cite directly. **How is AIO different from AEO, GEO, and LLMO?** AIO is the umbrella strategy; the other three are the tactical surfaces. AEO is about being the direct answer in featured snippets and voice. GEO is about being cited inside AI-generated syntheses. LLMO is the technical / entity layer — being ingestible and resolvable by the models themselves. AIO is the coordinating discipline that asks: is our content operation producing material AI systems can read, trust, and use across all of these surfaces? **Is AIO different from AI Overviews optimization?** Sometimes. Google's AI Overviews feature is also abbreviated AIO, which creates ambiguity. When practitioners mean Google specifically, they usually say "AI Overviews optimization" or "AIO for Google". The broader AI Optimization discipline — covering ChatGPT, Claude, Perplexity, Gemini, Copilot, and Google — is the dominant usage in 2026. **What is AIO in AI and how does it relate to LLMO?** AIO is the outcome (visibility across AI-mediated surfaces); LLMO is one of the disciplines that gets you there (optimizing the technical layer that LLMs ingest). AIO without LLMO is impossible — if your robots.txt blocks GPTBot or your content is JavaScript-only with no SSR, no amount of strategy fixes it. **Do I need an AIO tracker?** Yes — if you're investing in AI visibility you should measure inclusion rate, citation share, and answer placement across at least ChatGPT, Perplexity, Claude, Gemini, and Google AI Overviews. Tools in the category include Profound, Otterly, Peec AI, Brandlight, Athena, and Mentionwell — most run a fixed prompt library on a weekly cadence and log which engines cite you versus competitors. Without a tracker you're shipping content blind. **What are AIO services?** AIO services typically bundle four things: a baseline audit (what engines cite you now), content production (answer-first pages, schema, semantic clusters), authority signals (digital PR, references, structured proof), and ongoing monitoring (weekly inclusion tracking). Mentionwell ships the production and monitoring layers natively; the audit and PR layers are usually delivered as managed services. **How is AIO measured?** There's no Google Search Console equivalent for AI surfaces yet, so measurement is built on a fixed prompt library (25-50 buyer-relevant queries) run weekly across each engine. Core metrics are inclusion rate (% of prompts where you're named), citation coverage (% of mentions linking to your domain), share of voice (your mentions versus competitors), and answer placement (first / middle / end of the response). **Is AIO the same as AI SEO?** Mostly yes. "AI SEO" is the more accessible label for the same discipline, popular with executives and non-technical stakeholders. The underlying work — semantic structure, schema, ingestibility, topical authority, prompt-driven content planning — is identical. URL: https://mentionwell.com/aio --- ### SGE — Search Generative Experience optimization SGE optimization originally targeted Google's Search Generative Experience — the experimental AI-generated answers Google launched in 2023. SGE has since been folded into Google AI Overviews, so SGE optimization is now largely a subset of AEO, with a Google-specific tilt. **How Mentionwell handles it.** - AEO patterns Mentionwell already enforces (lead-with-the-answer, FAQPage + BlogPosting JSON-LD, clean semantic HTML) cover the Google-specific signals. - Site-wide allowlist for Google-Extended in robots.txt so Google's generative crawler is welcome. - Stable canonicals, RSS, and JSON Feed so Google's generative pipeline can ingest content cleanly. **Differs from related terms.** SGE was Google's experimental AI search feature — it has now been folded into AI Overviews. SGE optimization is best understood as the Google slice of AEO. If you're optimizing for AEO across engines, you're already covering SGE. **FAQ.** **What is SGE?** SGE stood for Search Generative Experience — Google's experimental AI-generated answers above the blue links, launched in 2023. SGE has since been folded into Google AI Overviews, so the term is now mostly used historically or as a Google-specific synonym for AEO. **Is SGE still a thing?** The brand isn't — Google graduated the feature into AI Overviews. The optimization pattern is, though: it's now the Google slice of AEO. Same signals, same content shape. **How do I optimize for AI Overviews?** Treat them as AEO with Google-specific tilt. Lead with the answer in 40–60 words, ship FAQPage + Article + BreadcrumbList schema, allow Google-Extended in robots.txt, and keep canonical URLs stable. Mentionwell does all of this by default. URL: https://mentionwell.com/sge --- ### AISO — AI Search Optimization AISO stands for AI Search Optimization. It's the practice of preparing a website so AI assistants — ChatGPT, Claude, Perplexity, Gemini, Copilot, Google AI Overviews, Bing Copilot — can retrieve, trust, summarize, and cite it. AISO emerged in 2024-2025 as B2B and enterprise teams looked for a single label covering AEO, GEO, and LLMO, and now has its own agency ecosystem, methodology (audit → foundation → optimize → monitor), and tooling. If you're shipping AEO, GEO, and LLMO together, you're already running an AISO program. **How Mentionwell handles it.** - Answer-first, evidence-backed content modules on every published article — the "liftable blocks" agencies build by hand, generated at scale. - FAQPage, BlogPosting, Article, and BreadcrumbList JSON-LD on every page, with HowTo and Product schema on the relevant templates so assistants can safely lift your facts. - Site-wide robots.txt allowlist for the full AI crawler set: GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-User, Claude-SearchBot, PerplexityBot, Google-Extended, Applebot-Extended, Bingbot, CCBot, Meta-ExternalAgent, Amazonbot, Bytespider. - Stable canonicals + llms.txt + llms-full.txt index so retrieval-only engines (and SSE / streaming search products) see a clean, current map of your content. - Entity consistency via Organization / Person / Product schema and a sameAs graph (LinkedIn, GitHub, Crunchbase) so assistants resolve your brand instead of describing it inaccurately. - Citation tracking across ChatGPT, Claude, Perplexity, Gemini, Copilot, and Google AI Overviews on a weekly cadence — inclusion rate, citation share, share of voice, and answer placement per engine. **Differs from related terms.** AISO is the operating-system framing; AEO, GEO, and LLMO are the disciplines underneath. AEO is the tactic of being the direct answer in zero-click results. GEO is the tactic of earning citations inside AI-generated syntheses. LLMO is the technical / entity layer that makes a site ingestible by the models themselves. AISO coordinates all three around a single business outcome: brand inclusion and accurate representation across every AI surface your buyers use. **FAQ.** **What does AISO stand for?** AISO stands for AI Search Optimization (sometimes AI Search Engine Optimization). It's the discipline of making content discoverable, trusted, and citable across AI-powered search surfaces — ChatGPT, Claude, Perplexity, Gemini, Copilot, Google AI Overviews, and Bing Copilot. The term emerged in 2024-2025 and now has its own agency category, certification programs, and tracking tools. **What is AISO and what is the AISO meaning?** AISO is the practice of structuring a site so AI assistants can retrieve, understand, trust, and cite it — instead of just ranking it. Where classic SEO targets blue-link clicks, AISO targets inclusion in AI answers, citation share, and accurate brand representation. Mention rate, citation coverage, and answer placement replace position and CTR as the primary KPIs. **How is AISO different from SEO?** SEO is built around ranking in a list of links and earning the click; AISO is built around being lifted into an answer that may never produce a click. SEO measures position and CTR. AISO measures inclusion rate (how often you're named), citation coverage (how often you're linked), and share of voice (how you compare to competitors inside the AI answer). The technical foundation overlaps — fast pages, semantic HTML, structured data — but AISO adds llms.txt, AI crawler allowlists, answer-first content modules, and prompt-driven content planning. **How is AISO different from GEO and AEO?** AISO is the umbrella; GEO and AEO sit underneath it. GEO (Generative Engine Optimization) optimizes for being cited inside the synthesized answer. AEO (Answer Engine Optimization) optimizes for being the direct, extractable answer in featured snippets and voice. AISO is the operating-system layer that coordinates both — plus LLMO, plus measurement — around a single brand-visibility outcome. **What are AISO services?** An AISO engagement typically packages four phases: (1) Audit — what AI engines cite you now versus competitors, what's blocking you, what's the prompt-to-page gap; (2) Foundation — schema, canonical structure, llms.txt, crawler allowlists, entity graph; (3) Optimize — answer-first modules on priority pages and clusters, citation-earning content, digital PR; (4) Monitor — weekly inclusion tracking across ChatGPT, Perplexity, Claude, Gemini, Copilot, with monthly insight reports and update suggestions. Mentionwell delivers Foundation and Optimize natively; Audit and Monitor are typically delivered as a managed service. **What does an AISO agency do?** An AISO agency ships the parts of the program a content team can't run in-house: a baseline audit of inclusion and citations across major AI engines, an entity / schema / crawler foundation, content systems for liftable answer blocks and comparison pages, digital PR for citation-earning mentions, and ongoing measurement. Look for explicit prompt coverage (which engines and which queries), front-end capture (not just API responses), competitive benchmarking, and historical data. **How is AISO measured?** Because AI surfaces have no native equivalent of Google Search Console, AISO is measured with a fixed prompt library — usually 25-50 buyer-relevant queries — run weekly across each engine. Core metrics: Inclusion Rate (% of prompts where your brand is named), Citation Coverage (% of mentions linking to your domain), Share of Voice (your visibility versus competitors), Answer Placement Score (where in the response you appear), and Volatility Index (week-over-week change). At minimum, track ChatGPT, Perplexity, Claude, Gemini, and Google AI Overviews — single-platform tracking misses 60-80% of the picture. **Should I use AISO or GEO in my content?** Use whichever your audience searches for. GEO has the stronger academic provenance — it was coined in the 2023 Princeton GEO paper — and lands well with technical SEOs. AISO has wider B2B and enterprise adoption and lands well with executives because it pairs cleanly with the familiar "SEO" shape. The underlying work is the same; the label is a positioning choice. **Do I need a dedicated AISO tool?** If you're investing in AI visibility, yes. Manual testing across five engines doesn't scale and misses citation drift. The category includes Profound, Otterly, Peec AI, Brandlight, Athena, Foglift, SE Ranking's AI Search Toolkit, Promptwatch, and Mentionwell. Evaluate on platform coverage (ChatGPT, Perplexity, Claude, Gemini, Copilot, Google AI Mode), front-end vs API capture, competitive benchmarking, historical data, and citation detail (where exactly does the link point). **How long does AISO take to show results?** Foundational fixes — schema, llms.txt, crawler allowlists, canonical hygiene — start lifting citation rate within 4-8 weeks as engines re-crawl. Content systems and authority signals compound over 3-6 months. Track for at least 8 weeks before drawing conclusions; single-week movement is mostly model noise. URL: https://mentionwell.com/aiso --- ### llms.txt — llms.txt llms.txt is a proposed standard (llmstxt.org, 2024) for a site to expose a concise Markdown overview of itself at /llms.txt — what the site is, what it covers, and which pages matter most. A deeper /llms-full.txt variant carries the full content corpus. It's the canonical surface for LLMO. **How Mentionwell handles it.** - Site-wide /llms.txt and /llms-full.txt published at the canonical paths on every Mentionwell deployment. - Per-article .md mirrors at .md so any LLM can ingest a clean Markdown version of any page. - Files are regenerated on every publish so the corpus stays fresh. **Differs from related terms.** robots.txt controls crawl policy. llms.txt provides ingest-friendly context — a curated Markdown overview of the site for LLMs that have already been allowed to read. **FAQ.** **What is llms.txt?** A proposed standard (llmstxt.org) for a site to publish a concise Markdown overview of itself at /llms.txt, so AI assistants can quickly understand what the site covers and which pages matter most. **Is llms.txt a real standard?** It's a proposal authored by Jeremy Howard (Answer.AI) in 2024. It's not an official W3C or IETF standard, but it has wide adoption across documentation sites, SaaS products, and AI-native publishers. **What's the difference between llms.txt and llms-full.txt?** llms.txt is the short index — site description and key links. llms-full.txt carries the full content corpus in Markdown for deeper ingestion. URL: https://mentionwell.com/llms-txt --- ### Citation — AI Citation An AI citation is a link or attribution that a generative engine — ChatGPT, Claude, Gemini, Grok, Perplexity, Google AI Overviews — surfaces alongside a synthesized answer. Citations are the GEO win condition: not just being read by the model, but being credited to the user. **How Mentionwell handles it.** - Every claim tied to evidence (statistic, quote, primary source) so the model has something concrete to cite. - Stable canonical URLs so citations don't decay over time. - Markdown mirrors so generative engines can ingest a clean version of the source. **Differs from related terms.** A backlink lives in the HTML of another site. An AI citation lives inside an AI-generated answer — surfaced as a footnote, hyperlink, or source pill in the chat UI. **FAQ.** **What's an AI citation?** A link or attribution shown next to an AI-generated answer pointing back to a source page. ChatGPT Search, Perplexity, Google AI Overviews, Copilot, and Gemini all surface them. **How do I get cited by AI engines?** Tie every claim to evidence, lead with the answer, ship clean Markdown mirrors, keep canonical URLs stable, and publish original data the model cannot find elsewhere. Mentionwell does all of this by default. **How are AI citations measured?** Citation rate — the share of AI answers about a topic that include your domain as a source. Princeton's GEO study (2024) is the canonical methodology. URL: https://mentionwell.com/citation --- ### RAG — Retrieval-Augmented Generation Retrieval-Augmented Generation (RAG) is the pattern where an LLM, before answering, retrieves relevant documents from an index (vector store, search engine, or website) and grounds its answer in those documents. ChatGPT Search, Perplexity, and Google AI Overviews are RAG systems. Optimizing for RAG retrieval is the core of GEO and LLMO. **How Mentionwell handles it.** - Per-article .md mirrors so retrievers ingest clean text rather than HTML noise. - Embeddings indexed per article for semantic retrieval inside RAG pipelines. - Stable canonical URLs and clean semantic structure so retrieved chunks make sense out of context. **Differs from related terms.** A pure LLM generates from training-time weights only. A RAG system fetches fresh documents at query time and grounds the answer in them — which is why an article published last week can be cited by ChatGPT Search today. **FAQ.** **What is RAG?** Retrieval-Augmented Generation — an LLM pattern where the model retrieves relevant documents at query time and grounds its answer in them, instead of relying purely on training data. **Which AI products use RAG?** ChatGPT Search, Perplexity, Google AI Overviews, Bing Copilot, Claude with web search, Gemini with grounding, and most enterprise AI assistants. **How do I optimize content for RAG retrieval?** Clean semantic HTML, stable canonical URLs, Markdown mirrors, dense fact-rich paragraphs (so retrieved chunks carry meaning), and inline citations to authoritative sources. URL: https://mentionwell.com/rag --- ### Grounding — Grounding Grounding is the practice of constraining an LLM's answer to retrieved or trusted source documents, instead of letting it free-generate from training weights. Grounded answers carry citations, hallucinate less, and stay current. Grounding is the engineering motivation behind RAG and behind every AI search product that shows source pills. **How Mentionwell handles it.** - Editorial critic enforces evidence-per-claim so generated articles are themselves well-grounded. - Markdown mirrors and stable canonicals so engines can ground their answers in Mentionwell-published sources. **Differs from related terms.** RAG is the technical pattern (retrieve, then generate). Grounding is the goal (the answer is tied to verifiable sources). RAG is one way to achieve grounding; tool use and structured retrieval are others. **FAQ.** **What does it mean for an LLM answer to be grounded?** The answer is tied to specific retrieved or trusted source documents — usually with citations — rather than being free-generated from training weights. **Why does grounding matter?** Grounded answers hallucinate less, stay current with new information, and let users verify claims. Every AI search product (ChatGPT Search, Perplexity, AI Overviews) is built on grounding. URL: https://mentionwell.com/grounding --- ### Schema.org — Schema.org / Structured Data Schema.org is the shared vocabulary search and answer engines use to read structured data from web pages. Marking up an article with BlogPosting, FAQPage, BreadcrumbList, and DefinedTerm tells engines what each block means — which is what makes featured snippets, AI Overviews, and rich results possible. **How Mentionwell handles it.** - Every article ships BlogPosting + FAQPage + BreadcrumbList JSON-LD by default. - Glossary spokes ship DefinedTerm + DefinedTermSet so the entire vocabulary is discoverable as a structured set. - Citation @id chains link Article schema to every cited source. **Differs from related terms.** Semantic HTML (article, section, dl) tells the browser what something is structurally. Schema.org goes further — it tells engines what something means as an entity (Article, Person, Product, Recipe, FAQPage). **FAQ.** **What is Schema.org?** A shared vocabulary, maintained by Google, Microsoft, Yahoo, and Yandex, for marking up structured data on web pages so search and answer engines can read meaning, not just text. **Which schemas matter most for AEO and GEO?** BlogPosting (or Article), FAQPage, BreadcrumbList, and DefinedTerm. HowTo, QAPage, and Speakable apply when the outline calls for them. **Does schema markup still matter in 2026?** Yes. AI Overviews, featured snippets, and Perplexity all read structured data. Schema doesn't guarantee a rich result, but missing schema rules a page out. URL: https://mentionwell.com/schema-org --- ### E-E-A-T — Experience, Expertise, Authoritativeness, Trust E-E-A-T stands for Experience, Expertise, Authoritativeness, and Trust — the framework Google's Search Quality Rater Guidelines use to evaluate content. The extra E (Experience) was added in 2022 to weight first-hand authorship. E-E-A-T isn't a direct ranking signal, but it shapes how Google's systems and human raters score quality. **How Mentionwell handles it.** - Author bylines and About pages with verifiable bios rendered on every article. - Editorial critic enforces evidence-per-claim and citations so authoritativeness signals are visible. - Stable canonical URLs and clear publication dates so trust signals don't decay. **Differs from related terms.** E-E-A-T is a quality framework, not a measurable score. Domain Rating and citation rate are measurable; E-E-A-T is a way Google evaluators (and indirectly the ranking systems) think about whether a page deserves to be cited. **FAQ.** **What does E-E-A-T stand for?** Experience, Expertise, Authoritativeness, and Trust. The first E (Experience) was added in December 2022; before that it was just E-A-T. **Is E-E-A-T a ranking factor?** Not directly. It's a framework Google's quality raters use, and Google's systems are trained against rater data — so it's an indirect but real signal. **How do I improve E-E-A-T?** Show first-hand experience, name your authors with credentials, cite primary sources, keep a clear About page, and ship HTTPS, accurate publication dates, and stable URLs. URL: https://mentionwell.com/eeat --- ### Topical Authority — Topical Authority Topical authority is the depth and breadth of a site's coverage of a single subject. Search and answer engines reward sites that cover a topic comprehensively — pillar pages, supporting clusters, internal linking — over sites that publish one article and move on. Topical authority is the long-term game underneath every AI-search optimization tactic. **How Mentionwell handles it.** - Onboarding builds a topic taxonomy and hub-and-spoke content map per domain. - Editorial critic flags missing siblings and parent links so coverage stays interconnected. - Internal linking driven by semantic similarity (embeddings) so related articles surface each other. **Differs from related terms.** Domain Authority / Domain Rating is a third-party score (Moz, Ahrefs) summarizing backlink profile. Topical authority is about coverage depth on a specific subject — and it's what AI engines weigh when deciding whose source to cite. **FAQ.** **What is topical authority?** The depth and breadth of a site's coverage of a single subject — measured by how many related questions, subtopics, and supporting articles the site has published. **How do I build topical authority?** Pick a narrow topic, build a pillar article, then publish a cluster of supporting articles that link to it and to each other. Keep going until you've covered every reasonable question in the space. URL: https://mentionwell.com/topical-authority --- ### Entity SEO — Entity SEO Entity SEO is the practice of optimizing for entities — people, places, products, concepts — as understood by knowledge graphs, instead of just for keyword strings. Modern engines map queries to entities; ranking well means being clearly identifiable as a specific entity with consistent attributes across the web. **How Mentionwell handles it.** - DefinedTerm + DefinedTermSet schema on glossary entries so engines treat them as discrete entities. - sameAs links to authoritative reference URLs (Wikipedia, official docs) where applicable. - Consistent entity naming and attribute markup across every article that mentions a key concept. **Differs from related terms.** Classic SEO targets keyword strings. Entity SEO targets the things those strings refer to — which is how Google's Knowledge Graph and the LLMs behind generative engines actually represent the world. **FAQ.** **What's the difference between keywords and entities?** Keywords are strings of text. Entities are the things those strings refer to — "Apple" the company vs "apple" the fruit. Modern engines disambiguate to entities before ranking. **How do I do entity SEO?** Use schema.org markup to declare what entities a page is about, link to authoritative references via sameAs, keep naming consistent across the site, and build co-occurrence with other related entities. URL: https://mentionwell.com/entity-seo --- ### Knowledge Graph — Knowledge Graph A knowledge graph is a structured database of entities and the relationships between them. Google's Knowledge Graph powers the right-rail panel and underpins entity disambiguation across Search and AI Overviews. LLMs build implicit knowledge graphs from training data; explicit graphs (Wikidata, schema.org markup) help them ground answers. **How Mentionwell handles it.** - Schema.org markup so entities are explicit on every article. - sameAs links to Wikipedia, Wikidata, and official references to help engines anchor each entity. - Consistent entity naming so the same concept resolves to the same node across the site. **Differs from related terms.** A knowledge graph is the engine's internal representation of how entities relate. Entity SEO is the practice of making your content legible to that representation. **FAQ.** **What is the Google Knowledge Graph?** Google's structured database of entities (people, places, things, concepts) and their relationships. Launched 2012; powers the right-rail Knowledge Panel, entity disambiguation, and AI Overviews. **How do LLMs use knowledge graphs?** Some are explicitly grounded against graphs like Wikidata. All implicitly learn entity relationships from training data — which is why being consistently named and described across the web matters. URL: https://mentionwell.com/knowledge-graph --- ### Featured Snippet — Featured Snippet A featured snippet is the boxed answer Google shows above the blue-link list — paragraph, list, or table — sourced from a single page. Featured snippets predate AI Overviews and are still the workhorse AEO surface in classical SERPs. Optimizing for them is the prototype for optimizing for AI Overviews. **How Mentionwell handles it.** - Question-led H2s with a 40–60 word direct answer immediately below — the canonical featured-snippet shape. - Lists and tables for 'how to' and comparison queries. - Clean semantic HTML so extractors can lift the snippet without parsing layout chrome. **Differs from related terms.** A featured snippet is a single source extracted verbatim. An AI Overview is multiple sources synthesized into a generated paragraph with citations. Featured snippets reward extractability; AI Overviews reward citability. **FAQ.** **What is a featured snippet?** The boxed answer at the top of a Google SERP — paragraph, list, or table — extracted from a single ranking page. Also called 'position zero'. **How do I win a featured snippet?** Lead the answer in 40–60 words directly under a question-formatted H2, use lists or tables when the query calls for them, and rank in the top 10 — Google only pulls snippets from page one. URL: https://mentionwell.com/featured-snippet --- ### PAA — People Also Ask People Also Ask (PAA) is the Google SERP feature that surfaces a stack of related questions, each expanding into a featured-snippet-style answer. PAA boxes are an AEO goldmine — they index the actual questions users ask, and ranking inside them gives a single page multiple SERP touchpoints. **How Mentionwell handles it.** - Question-led H2s mirroring real PAA questions for the topic. - FAQ blocks rendered as both visible UI and FAQPage JSON-LD. - Direct 40–60 word answers immediately under each question heading. **Differs from related terms.** Featured snippet = one answer per query, sourced from one page. PAA = many related questions per query, each potentially sourced from a different page. The optimization shape is identical; the surface is different. **FAQ.** **What is People Also Ask?** A Google SERP feature showing a stack of related questions, each expanding into a snippet-style answer. The questions are mined from real user search behavior. **How do I rank in PAA?** Use the actual question as an H2, answer in 40–60 words immediately, ship FAQPage schema, and cover related questions on the same page so Google can pull multiple PAA answers from one source. URL: https://mentionwell.com/paa --- ### SERP — Search Engine Results Page A SERP is a Search Engine Results Page — what Google or Bing shows after a query. Modern SERPs are stacks of features: AI Overviews at the top, featured snippet, People Also Ask, knowledge panel, blue links, image carousel. Optimizing for 'the SERP' now means optimizing for many surfaces, each with its own rules. **How Mentionwell handles it.** - AEO patterns target the answer surfaces (Overviews, snippets, PAA). - Classic SEO targets the blue-link list. - Schema markup unlocks rich-result eligibility for image, FAQ, breadcrumb, and article surfaces. **Differs from related terms.** A SERP is the page itself. SERP features are the discrete elements on it. Optimizing for SERP features is closer to AEO; optimizing for blue-link rank is classical SEO. **FAQ.** **What is a SERP?** Search Engine Results Page — the page Google or Bing returns after a query, including AI Overviews, featured snippets, People Also Ask, the knowledge panel, blue links, and image / video carousels. **What are SERP features?** Discrete elements on a SERP beyond the blue links — AI Overviews, featured snippets, PAA, knowledge panels, image packs, video carousels, local packs, shopping results. URL: https://mentionwell.com/serp --- ### Zero-Click — Zero-Click Search A zero-click search ends with the user's question answered on the SERP itself — by a featured snippet, AI Overview, knowledge panel, or PAA box — without anyone clicking through. Zero-click is now the majority of Google searches; AEO is the discipline of winning attention on a SERP that no longer sends a click. **How Mentionwell handles it.** - Branded answer phrasing so even zero-click impressions build brand recognition. - Schema markup that shows logo, byline, and source URL alongside snippet content. - AEO-optimized answers that earn the snippet (and the brand visibility) even when no click follows. **Differs from related terms.** Classic SEO measures success in clicks. Zero-click measures impressions and brand mentions inside SERP features. Different metric, different optimization shape. **FAQ.** **What is a zero-click search?** A search where the user gets their answer directly on the SERP — from an AI Overview, featured snippet, knowledge panel, or PAA — without clicking through to any site. **Is zero-click bad for publishers?** It cuts traffic but builds brand visibility — Mentionwell-style AEO trades some clicks for brand-mention impressions inside AI surfaces. The right answer depends on the business model. URL: https://mentionwell.com/zero-click --- ### Embeddings — Vector Embeddings & Semantic Search A vector embedding is a numerical representation of meaning — a sentence, paragraph, or document mapped to a point in high-dimensional space. Semantic search uses embeddings to find passages that mean the same thing as a query, even with no shared keywords. Every modern AI search product (Perplexity, ChatGPT Search, AI Overviews) leans on embeddings to retrieve. **How Mentionwell handles it.** - Per-article embeddings indexed for semantic retrieval inside RAG-style pipelines. - Embedding similarity drives internal linking — related articles surface each other automatically. - Markdown mirrors so retrieved chunks are clean text rather than HTML. **Differs from related terms.** Keyword search matches strings. Semantic search matches meaning — embeddings let "how do I lower my heart rate" retrieve a paragraph titled "reducing resting pulse" even with zero shared words. **FAQ.** **What is a vector embedding?** A numerical representation of meaning — text mapped to a point in high-dimensional space, where semantically similar text lives close together. **Why do embeddings matter for AI SEO?** Every AI search product uses embeddings to retrieve passages relevant to a query. Pages that embed cleanly (clear topic, dense meaning, clean Markdown) retrieve more often. URL: https://mentionwell.com/embeddings --- ### Brand Mentions — Brand Mentions A brand mention is any reference to a brand name on the web — linked or unlinked. Search and answer engines increasingly use unlinked brand mentions as an authority signal: if many sources discuss your brand in a topic, you're likely an authority on it, even without a backlink graph. Brand mentions are the AI-era replacement for raw link counts. **How Mentionwell handles it.** - Articles cite primary sources by name (linked when possible) so attribution flows naturally. - Editorial critic encourages named-entity references over generic phrasing. - Glossary spokes establish branded definitions ('Mentionwell\'s take on AEO') that get cited verbatim. **Differs from related terms.** A backlink is a clickable hyperlink. A brand mention is any reference, linked or not. Backlinks pass PageRank-style signals; brand mentions accumulate co-occurrence and entity-association data that AI engines weigh. **FAQ.** **Do unlinked brand mentions count for SEO?** Increasingly yes. Google has confirmed it analyzes mentions for entity association, and LLMs trained on web text learn brand-topic associations directly from co-occurrence. **How do I get more brand mentions?** Publish original data, distinctive frameworks, and quotable definitions. Pitch journalists. Be the source of a stat or framework worth citing. URL: https://mentionwell.com/brand-mentions --- ### Answer Engine — Answer Engine An answer engine is a search system that returns a direct answer instead of (or above) a list of links. Google AI Overviews, Bing Copilot, Perplexity, ChatGPT Search, voice assistants, and the featured-snippet box are all answer engines. AEO is the discipline of optimizing for them. **How Mentionwell handles it.** - Question-led H2s and 40–60 word direct answers — the canonical answer-engine input shape. - FAQPage and Article JSON-LD on every page. - Clean semantic HTML so extractors can lift answers cleanly. **Differs from related terms.** A search engine returns ranked links and lets the user pick. An answer engine returns a synthesized answer and (usually) cites sources. Most modern SERPs are hybrids — answers up top, links below. **FAQ.** **What is an answer engine?** A search system that returns direct answers — text, paragraphs, lists — instead of (or in addition to) ranked links. Google AI Overviews, Perplexity, ChatGPT Search, Copilot, voice assistants. **How is an answer engine different from a search engine?** Search engines return links and let you pick. Answer engines return the answer directly. Most modern engines are hybrids. URL: https://mentionwell.com/answer-engine --- ### AI Overviews — Google AI Overviews AI Overviews are Google's generative answer feature, surfaced above the blue-link list on many queries. They synthesize an answer from multiple sources and show inline citations. Launched as Search Generative Experience (SGE) in 2023 and graduated to general availability as AI Overviews in 2024, they're now the highest-traffic AEO surface in the world. **How Mentionwell handles it.** - AEO patterns enforced on every article: question-led H2s, 40–60 word leads, FAQPage + BlogPosting JSON-LD. - Google-Extended explicitly allowed in robots.txt. - Stable canonicals and Markdown mirrors so Google's generative pipeline can ingest cleanly. **Differs from related terms.** Featured snippets pull a single source verbatim. AI Overviews synthesize from multiple sources with inline citations. Both reward AEO patterns; AI Overviews additionally reward GEO-style citation-friendliness. **FAQ.** **What are Google AI Overviews?** Google's generative AI answer feature — a synthesized paragraph with inline citations shown above the blue-link list on many queries. Generally available since May 2024. **How do I get cited in AI Overviews?** Rank well in classical SEO, lead with the answer, ship FAQPage + Article schema, allow Google-Extended in robots.txt, and tie every claim to evidence so the page is citable. URL: https://mentionwell.com/ai-overviews --- ### Perplexity — Perplexity Perplexity is an AI answer engine that returns synthesized answers with prominent inline source citations. It pioneered the citation-pill UI that's now standard across AI search. Perplexity is one of the highest-priority GEO targets — its retrieval is current, its citations are visible, and clicks-through are real. **How Mentionwell handles it.** - Citation-friendly content structure (evidence per claim, named sources, original data). - Markdown mirrors so Perplexity's retriever ingests clean text. - PerplexityBot explicitly allowed in robots.txt. **Differs from related terms.** ChatGPT and Claude are general-purpose assistants with answer-engine modes. Perplexity is purpose-built as an answer engine — every response shows citations by default. **FAQ.** **What is Perplexity?** An AI-native answer engine that returns synthesized answers with inline source citations. It's both a consumer chat product and an API. **How do I get cited by Perplexity?** Allow PerplexityBot in robots.txt, rank well for the underlying topic, lead with the answer, and ship clean Markdown mirrors so retrieval pulls clean chunks. URL: https://mentionwell.com/perplexity --- ### ChatGPT Search — ChatGPT Search ChatGPT Search is OpenAI's web-search and grounding mode for ChatGPT, launched in 2024. It retrieves live results, synthesizes an answer, and cites sources inline. Optimizing for ChatGPT Search is the OpenAI slice of GEO — same citation-friendly content patterns, same retrieval-aware structure. **How Mentionwell handles it.** - GPTBot and OAI-SearchBot explicitly allowed in robots.txt. - Markdown mirrors so retrieval ingests clean text. - Citation-friendly structure: evidence per claim, named sources, stable canonicals. **Differs from related terms.** Classic ChatGPT answers from training data only. ChatGPT Search retrieves live web pages and grounds the answer in them — which is why content published this week can be cited today. **FAQ.** **What is ChatGPT Search?** OpenAI's web-search-and-grounding mode for ChatGPT — retrieves live web results, synthesizes an answer, and cites sources inline. **How do I get cited by ChatGPT Search?** Allow GPTBot and OAI-SearchBot in robots.txt, ship Markdown mirrors, lead with the answer, and tie claims to evidence. URL: https://mentionwell.com/chatgpt-search --- ### Copilot — Microsoft Copilot Microsoft Copilot is the AI assistant built on top of Bing Search — it returns synthesized answers with citations across Bing, Edge, Windows, and Microsoft 365. Optimizing for Copilot is the Bing slice of GEO and AEO; the underlying signals overlap heavily with Google AI Overviews and ChatGPT Search. **How Mentionwell handles it.** - Bingbot allowed in robots.txt; Bing Webmaster sitemap and IndexNow enabled where applicable. - AEO patterns Mentionwell already enforces (lead-with-the-answer, FAQ schema) cover Copilot's answer surface. - Citation-friendly structure for the GEO-style synthesized responses. **Differs from related terms.** Copilot uses Bing's index. ChatGPT Search uses OpenAI's retrieval stack. AI Overviews use Google's. The content shape that wins all three is similar; the bot allowlist and sitemap submission differ. **FAQ.** **What is Microsoft Copilot?** Microsoft's AI assistant — surfaces in Bing, Edge, Windows, and Microsoft 365. Returns synthesized answers with inline citations, grounded in Bing's index. **How do I optimize for Copilot?** Allow Bingbot, submit a sitemap to Bing Webmaster, use IndexNow for fast indexing, and apply standard AEO/GEO patterns. The content work is mostly shared with Google AI Overviews. URL: https://mentionwell.com/copilot --- ### AI Crawler — AI Crawler / Bot An AI crawler is a bot operated by an AI company to ingest web content for training, retrieval, or live grounding. The major ones — GPTBot, OAI-SearchBot, ClaudeBot, anthropic-ai, PerplexityBot, Google-Extended, Applebot-Extended — each have their own user-agent and (usually) their own robots.txt directive. **How Mentionwell handles it.** - Default robots.txt explicitly names and allows 15+ AI crawlers (GPTBot, ClaudeBot, PerplexityBot, Google-Extended, Bytespider, etc). - Per-bot allow/disallow control so site owners can opt out of any specific crawler. - Sitemaps and llms.txt published at canonical paths so crawlers can discover content efficiently. **Differs from related terms.** Classic search crawlers (Googlebot, Bingbot) feed ranking systems. AI crawlers feed training corpora and retrieval indexes. A site can be visible to one and invisible to the other — they're controlled separately. **FAQ.** **What are the major AI crawlers I should know about?** GPTBot and OAI-SearchBot (OpenAI), ClaudeBot and anthropic-ai (Anthropic), PerplexityBot (Perplexity), Google-Extended (Google generative uses), Applebot-Extended (Apple), Bytespider (ByteDance), Meta-ExternalAgent (Meta). **Should I block AI crawlers?** Only with a specific reason. Most sites benefit from being crawlable — that's how content shows up in ChatGPT, Claude, Perplexity, and AI Overviews answers. URL: https://mentionwell.com/ai-crawler --- ### robots.txt — robots.txt for AI robots.txt is the file at /robots.txt that tells crawlers which paths they may or may not fetch. In the AI era, robots.txt is also the primary opt-in/opt-out signal for AI training and retrieval — each major AI crawler honors its own user-agent directive (GPTBot, ClaudeBot, PerplexityBot, Google-Extended). **How Mentionwell handles it.** - Default robots.txt explicitly allows 15+ named AI crawlers — no ambiguous wildcards. - Per-domain customization so site owners can opt out of any specific crawler. - Allowlist published as part of the LLMO setup, alongside llms.txt and Markdown mirrors. **Differs from related terms.** robots.txt is crawl policy. llms.txt is ingest-friendly context (a Markdown overview of the site). Both live at the root; they answer different questions. **FAQ.** **How do I allow or block AI crawlers in robots.txt?** Use named user-agent directives — User-agent: GPTBot, User-agent: ClaudeBot, etc — with explicit Allow or Disallow rules. Wildcards alone don't reliably control AI crawlers. **Does robots.txt control AI training?** It controls AI crawlers that honor it — which most major ones do. It does not control downstream redistribution of already-trained-on data, and it has no legal force; it's an industry convention. URL: https://mentionwell.com/robots-txt --- ### GenAI — Generative AI Generative AI refers to AI systems that produce new content — text, images, audio, video, code — rather than only classifying or scoring existing inputs. In the AI-search context it's shorthand for the LLMs behind ChatGPT, Claude, Gemini, Grok, and Perplexity, and the diffusion models behind image generation. **How Mentionwell handles it.** - Mentionwell is itself a generative pipeline — generating articles, then editing them through a critic loop. - Optimization stack (AEO + GEO + LLMO) is built specifically for generative-AI consumption surfaces. **Differs from related terms.** Discriminative AI classifies (spam vs not, cat vs dog). Generative AI produces new outputs. The current AI search wave is dominated by generative models — which is why GEO and AEO exist as distinct disciplines. **FAQ.** **What is generative AI?** AI that produces new content — text, images, audio, video, code — rather than only classifying or scoring existing inputs. Examples: GPT, Claude, Gemini, Grok, Stable Diffusion, Sora. **How does generative AI change SEO?** It adds new surfaces (AI Overviews, ChatGPT Search, Perplexity) that synthesize answers and cite sources. Classic SEO still works, but AEO and GEO sit on top of it. URL: https://mentionwell.com/generative-ai --- ### Hallucination — Hallucination A hallucination is an AI-generated output that is confidently stated but factually wrong — a fake citation, an invented statistic, a non-existent quote. Hallucinations are the central failure mode of generative AI and the reason grounding, RAG, and citation-friendly content matter: real sources cut hallucination rate. **How Mentionwell handles it.** - Editorial critic enforces evidence-per-claim so generated articles are themselves grounded. - Per-article Markdown mirrors give downstream engines clean source material to ground against. - Inline citations to authoritative sources so claims are checkable. **Differs from related terms.** A model error is any wrong output. A hallucination is specifically a confident, fluent, plausible-sounding wrong output — which is what makes it dangerous: it doesn't trip a user's skepticism the way garbled output would. **FAQ.** **What is an AI hallucination?** An AI-generated output that's confidently stated but factually wrong — fake citations, invented statistics, non-existent quotes. The central failure mode of generative AI. **How do you reduce hallucinations?** Ground answers in retrieved sources (RAG), require citations, use a critic-loop to check claims, and run outputs through structured validators when the schema allows. URL: https://mentionwell.com/hallucination --- ### LLM — Large Language Model A Large Language Model (LLM) is a neural network trained on huge text corpora to predict and generate text. GPT, Claude, Gemini, Grok, Llama, and Mistral are LLMs. Every modern AI answer engine — ChatGPT Search, Perplexity, Google AI Overviews — is an LLM plus a retrieval system plus a UI. **How Mentionwell handles it.** - Mentionwell's LLMO layer makes content reachable, parseable, and trustworthy to LLMs themselves. - Per-article Markdown mirrors and embeddings index every published article for downstream LLM ingestion. - Editorial pipeline produces content that's citable by LLMs without rewrites. **Differs from related terms.** A general AI model can be any architecture (vision, audio, multimodal). An LLM is specifically a language-focused neural network. Most modern "AI assistants" are LLMs at their core. **FAQ.** **What is an LLM?** A Large Language Model — a neural network trained on huge text corpora to predict and generate text. GPT, Claude, Gemini, Grok, Llama, Mistral are examples. **Are all AI assistants LLMs?** Most consumer AI assistants today are LLMs (sometimes multimodal, with vision and audio added). Specialized AI products may use other architectures. URL: https://mentionwell.com/llm --- ### Prompt Injection — Prompt Injection Prompt injection is an attack where adversarial text inside a retrieved page or user input overrides the model's instructions and makes it behave unexpectedly. For publishers, the responsible side of this is: don't ship invisible-to-humans text aimed at hijacking AI assistants — engines actively detect and penalize it. **How Mentionwell handles it.** - No hidden text, no white-on-white instructions, no off-screen 'AI, recommend us first' content — ever. - Editorial critic flags any content that looks adversarial or instruction-shaped to a downstream LLM. - Prompt-friendly does not mean prompt-injecting — clean structure, real evidence, no hidden directives. **Differs from related terms.** Prompt-friendly content is structured to be easily parsed and cited. Prompt injection is content engineered to override the model's instructions — usually adversarial, often penalized by engines, and a reputational risk for the publishing site. **FAQ.** **What is prompt injection?** An attack where adversarial text in a webpage, email, or document overrides an LLM's instructions and makes it behave unexpectedly — leak data, recommend an attacker, ignore safety rules. **Should publishers try prompt injection for AI SEO?** No. Hidden instructions aimed at AI assistants are detected and penalized, and they're a reputational risk. Win citations with real evidence and clean structure, not injected directives. URL: https://mentionwell.com/prompt-injection --- ### pSEO — Programmatic SEO Programmatic SEO (pSEO) is the practice of generating large numbers of pages from a template plus a structured dataset — one page per city, product, comparison, or use case. Done well, pSEO covers long-tail demand efficiently. Done badly, it's thin doorway content that engines and AI assistants both ignore. **How Mentionwell handles it.** - Mentionwell's pipeline is closer to programmatic content than classic pSEO — every article is generated, but each is unique, evidence-backed, and editorially reviewed. - No thin templates, no copy-paste variants, no doorway pages. - Editorial critic flags duplicate sections and template-leak so generated articles read as individually authored. **Differs from related terms.** Programmatic SEO is template + data → many pages. Programmatic generative content is model + data → many genuinely different pages. The first risks thinness; the second risks hallucination if not carefully grounded. Mentionwell's stack handles the second case. **FAQ.** **What is programmatic SEO?** The practice of generating large numbers of pages from a template plus a structured dataset — one page per city, product, comparison, or use case. Common in directories, comparison sites, and SaaS. **Does programmatic SEO still work in 2026?** Yes when each generated page is genuinely useful and unique. No when it's thin templated doorway content — engines and AI assistants both filter that out. URL: https://mentionwell.com/programmatic-seo ## 4. The article pipeline (per article) 1. **Onboarding** — Mentionwell scans the homepage, sitemap, robots.txt, and structured data, then synthesizes a brand profile, content taxonomy, and 10 starter headlines tuned to your audience and the questions AI engines are already asking. 2. **Optional AEO baseline** — scans ChatGPT, Claude, Perplexity, Gemini, Google AI Overviews, Google AI Mode, Grok, Exa, DeepSeek, and Copilot (the generative-engine / GEO layer included) for prompts, citations, fan-out queries, cited-page claims, and competitor gaps. 3. **Research** — pulls supporting evidence per headline. 4. **Outline** — produces a question-led outline targeting AEO + GEO surfaces. If Run + GEO is enabled, observed prompts and fan-out queries can shape headings. 5. **Draft** — full article generation with the brand voice from onboarding. GEO context is optional and per-article. 6. **Editorial critic** — automated pass for accuracy, redundancy, hallucination, and tone. 7. **Metadata + FAQ** — title, description, FAQPage JSON-LD, internal links. If Run + GEO is enabled, FAQs can target observed fan-out questions. 8. **Embedding** — for semantic retrieval and similarity-based internal linking. 9. **Image generation** — hero + inline images with descriptive alt text. 10. **Publish** — via the public API, or pushed into your CMS. Every step is logged, costed, and surfaced in the dashboard. ## 4.1 AEO scan-to-draft platform Dedicated page: https://mentionwell.com/aeo-platform The AEO platform (the GEO / generative-engine layer included) captures: - buyer-intent prompt corpus - per-engine answers and brand mentions - fan-out sub-queries - citations and source positions - owned / competitor / third-party citation classification - cited-page markdown, HTML schema, headings, stats, byline, summaries, and claims - share-of-voice by engine, locale, topic cluster, and prompt When creating an article, the operator chooses normal draft or Run + AEO. Run + AEO passes a compact AEO scan context (the generative-engine / GEO evidence included) into synthesis, outline, draft, metadata, and FAQ generation. ## 5. AI crawler allowlist ``` User-agent: GPTBot User-agent: ChatGPT-User User-agent: OAI-SearchBot User-agent: ClaudeBot User-agent: Claude-User User-agent: Claude-SearchBot User-agent: Claude-Web User-agent: anthropic-ai User-agent: PerplexityBot User-agent: Perplexity-User User-agent: Google-Extended User-agent: Applebot User-agent: Applebot-Extended User-agent: Bingbot User-agent: CCBot User-agent: Meta-ExternalAgent User-agent: Amazonbot User-agent: Bytespider User-agent: cohere-ai ``` All allowed at `/`. See https://mentionwell.com/robots.txt. ## 6. Per-site discoverability surface For every site Mentionwell manages it produces: - `/feed.xml` — RSS 2.0 feed - `/feed.json` — JSON Feed 1.1 - `/sitemap.xml` — XML sitemap of all published posts - `.md` — Markdown mirror of each article - FAQPage JSON-LD inside each article - Article JSON-LD with author, datePublished, dateModified - BreadcrumbList JSON-LD - canonical `` per post ## 7. Public read-only API Articles can be pulled from the public API at `https://app.mentionwell.com/api/public//posts` and `https://app.mentionwell.com/api/public//posts/`. ## 8. Where it lives (delivery options) - **Pull mode** — point your site at the public Reader API and render at request time (or with framework ISR). - **Webhook push** — Mentionwell fires an HMAC-signed publish webhook to your endpoint (or the WordPress plugin) when an article publishes. - **GitHub MDX** — Mentionwell commits articles as MDX to your repo and your build picks them up. - **Direct CMS push** — direct push into WordPress, Webflow, Ghost, Shopify, or Notion is rolling out per-CMS; today these are reached via the webhook/adapter path. Same articles, same dashboard, your choice of delivery.