DEV Community: KDUAI

AI Visibility Glossary: 18 Terms in Plain English

KDUAI — Fri, 21 Aug 2026 10:57:19 +0000

AI visibility glossary: 18 terms in plain English

17 August 2026 · AI Visibility Index

Why this exists: the AI-search industry sprouts jargon faster than it sprouts standards. Half of it describes real mechanics; half is marketing fog. Here's the vocabulary you'll meet, in plain English, with an honest note on what actually matters.

The big ideas

AI visibility — How often AI engines name your business when people ask buying questions. Measured as a rate across many prompt runs, not a single ranking. The metric that matters most.

GEO (Generative Engine Optimisation) — The practice of improving how generative AI engines represent and recommend you. The AI-era sibling of SEO.

AEO (Answer Engine Optimisation) — Older term for the same idea, from the featured-snippet era. Now mostly folded into GEO.

Share of model — Of all the times an AI recommends someone in your category, the share that names you. Like share of voice, for AI answers.

Entity — A thing the model recognises as real and distinct: a business with a name, place, and consistent identity. Becoming a clean entity is half the battle — models recommend things, not strings of text.

Citation — When an AI answer links to or names a source. Being cited (as a source) and being recommended (as an option) are different wins; you want both.

The machinery

AI crawler — The bot that fetches your pages for an AI company. The ones to know: GPTBot (OpenAI training), OAI-SearchBot (ChatGPT search citations), ChatGPT-User (live browsing), PerplexityBot, ClaudeBot, Google-Extended. Each can be allowed or blocked separately in robots.txt — our allow-list guide has the config.

Retrieval — The step where the AI fetches current information before answering. If retrieval can't find you, no amount of training-data presence saves you.

RAG (Retrieval-Augmented Generation) — The architecture behind search-enabled AI: retrieve documents, then generate an answer grounded in them. Why "being findable" and "being quotable" are two separate problems.

Index — The database of crawled pages an engine searches. Google AI uses Google's index; ChatGPT leans on Bing's; Perplexity runs its own.

Grounding — Anchoring an answer in retrieved sources rather than the model's memory. Grounded answers are the ones that can name you correctly.

Hallucination — When a model states something false with confidence. Clean entities and consistent data across the web are the best defence against hallucinated claims about your business.

The formats

llms.txt — A proposed briefing file at your site root telling AI models what you are and what matters. Plain-English guide here.

JSON-LD — The script format for schema markup. Machine-readable identity for your business — with examples.

Schema / structured data — Standardised tags (schema.org) describing businesses, articles, FAQs. The vocabulary JSON-LD carries.

Answer-first content — Pages that lead with a 75–150 word direct answer before the detail. The format AI engines extract most reliably.

Prompt tracking — Running fixed sets of buyer questions against AI engines over time to measure visibility. What our data engine does nightly.

Frequently asked questions

Is GEO just SEO with a new name?

They overlap heavily — good structure, clear entities, quality content help both. The differences are in retrieval sources (Bing matters again), output format (sentences, not links), and measurement (visibility rates, not positions).

Which single term should I care about most?

AI visibility rate — the share of relevant buyer questions where you get named. Everything else is a means to move that number.

Are "AI SEO tools" that promise rankings legit?

Be sceptical of anyone promising fixed positions in AI answers — answers are regenerated every run and vary. Legitimate tools measure patterns across many runs and show their evidence. That's the standard to hold vendors to, us included.

Originally published at the AI Visibility Index.

What is AI visibility? A plain-English guide for local businesses

KDUAI — Sun, 16 Aug 2026 15:43:38 +0000

What is AI visibility? A plain-English guide for local businesses

16 August 2026 · AI Visibility Index

Direct answer: AI visibility is the likelihood that an AI search engine names your business when a customer asks for a recommendation in your category. It's measured as a rate — out of many real buyer questions ("best plumber in Denver"), in how many answers does the AI mention you? Unlike a Google ranking, there's no fixed position: answers are generated fresh each time, so visibility is a pattern across many runs, not a single slot.

For twenty years, "being found online" meant ranking on a results page. AI search replaces the results page with a sentence. When someone asks ChatGPT "who should I hire to fix a burst pipe in Denver?", the model names a handful of businesses — and everyone else simply doesn't exist in that conversation.

How AI visibility is measured

The honest way to measure it:

Write real buyer questions for your category and city — 15–25 of them, the way customers actually ask.
Run each one multiple times per engine (ChatGPT, Google AI, Perplexity), because answers vary run to run.
Score the pattern: in what share of answers were you named? Who was named instead? What sources were cited?

The output is an AI visibility rate — "named in 2 of 20 relevant searches" — plus the competitor set the AI prefers, and the sources it trusts. A single screenshot of one prompt proves nothing either way.

How it differs from SEO

	SEO	AI visibility
Output	Ranked list of links	Generated sentence naming options
Positions	Stable, tracked daily	Regenerated every run
Index	Google	Google + Bing + engine-specific retrieval
Fixes	Keywords, links, technical	Crawler access, entity signals, citable answers
Measurement	Rank trackers	Repeated prompt testing

Good SEO helps — Google's AI features use Google's index — but it's not sufficient. ChatGPT search leans on Bing; Perplexity retrieves from its own stack. Plenty of #1-ranked businesses are absent from AI answers, and some lower-ranked ones get named because their entity signals and answer-first pages are cleaner.

The three levers

Everything that moves AI visibility falls into three buckets:

Technical readiness — can the crawlers (GPTBot, PerplexityBot, Bingbot…) actually fetch your pages? See our robots.txt allow-list guide.
Entity strength — can the model confirm you're a specific, real business? Consistent name/address/phone, LocalBusiness schema, Google Business Profile, directory listings.
Answer quality — does any page give the model a short, sourced, extractable answer? Lead with 75–150 words of direct answer, then evidence.

Frequently asked questions

Can I pay to appear in AI answers?

Not directly. There's no ad slot inside a ChatGPT or Perplexity recommendation. What you can do is remove the blockers and supply better, more citable material than competitors.

How long does it take to improve?

Technical fixes (robots.txt, schema, indexation) can show effects within weeks as engines recrawl. Entity-building (directories, reviews, citations) compounds over months. It's a flywheel, not a switch.

Is AI visibility just a fad?

AI answers are increasingly the first step in local purchase journeys — Google's own AI Overviews now sit above the classic results page. Whether the share is 10% or 40% for your category today, the direction is one-way, and the fixes overlap heavily with good web practice anyway.

Originally published at the AI Visibility Index. Free research by the makers of aicantfindme.com — find out why AI search isn't recommending your business.

robots.txt for AI crawlers — the complete allow-list guide

KDUAI — Sun, 16 Aug 2026 15:43:35 +0000

robots.txt for AI crawlers: the complete allow-list guide

16 August 2026 · AI Visibility Index

Direct answer: To let AI search engines read and cite your site, your robots.txt must not block their crawlers. The critical ones are GPTBot and OAI-SearchBot (ChatGPT), PerplexityBot (Perplexity), ClaudeBot (Claude), Google-Extended (Google AI training), plus Bingbot — because ChatGPT's search relies heavily on Bing's index. Add explicit Allow: / rules for each, then verify by fetching your robots.txt in a browser.

If an AI crawler can't fetch your pages, the model learns about you from third-party sources — review sites, directories, competitors' comparisons — or from nothing. This is the single most common technical blocker we see, and it takes two minutes to check.

The crawlers that matter

Crawler	Belongs to	Blocks what if disallowed
GPTBot	OpenAI	ChatGPT training data
OAI-SearchBot	OpenAI	ChatGPT search citations
ChatGPT-User	OpenAI	Live browsing when users ask
PerplexityBot	Perplexity	Perplexity answers and citations
ClaudeBot	Anthropic	Claude's knowledge of your site
Google-Extended	Google	Gemini/AI training (separate from Search)
Bingbot	Microsoft	Bing index — feeds ChatGPT search and Copilot

A subtlety worth knowing: GPTBot and OAI-SearchBot are different gates. Blocking GPTBot keeps you out of future training data; blocking OAI-SearchBot keeps you out of cited answers today. If you want AI search traffic, the second one is the one that matters most.

Copy-paste allow-list

# AI search crawlers - allow
User-agent: GPTBot
Allow: /

User-agent: OAI-SearchBot
Allow: /

User-agent: ChatGPT-User
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: ClaudeBot
Allow: /

User-agent: Google-Extended
Allow: /

# Everyone else
User-agent: *
Allow: /
Disallow: /api/
Disallow: /admin/

Sitemap: https://yoursite.com/sitemap.xml

Keep the Disallow lines for genuine private areas only. Never put Disallow: / under User-agent: * — it blocks everything, including the AI crawlers you just allowed above (specific rules win, but a global block under * combined with missing specific rules is the classic failure).

Three common ways sites accidentally block AI

A managed host's default robots.txt. Some website builders ship a restrictive default. Check yours even if you never wrote one.
A CDN/WAF bot rule. Cloudflare and similar services can challenge bots before robots.txt is even read. If your security settings block "AI bots" as a category, robots.txt won't save you.
A blanket Disallow: / left over from a staging migration. Embarrassingly common.

How to verify it worked

Open https://yoursite.com/robots.txt in a browser — read what the world sees.
Confirm your site is in Bing's index: search site:yoursite.com on Bing. No results = ChatGPT search can't find you either. (We built a free index check for exactly this.)
Spot-check with a live prompt: ask ChatGPT "what does [yoursite.com] do?" If it can describe your services, crawling is working.

Frequently asked questions

Does allowing AI crawlers hurt my Google rankings?

No. robots.txt for AI crawlers is independent of Googlebot and Google Search. Google-Extended only governs AI training usage — Google states it does not affect Search inclusion or ranking.

Should I block AI crawlers to protect my content?

That's a legitimate choice for some publishers, but it trades away discoverability: engines can't recommend what they can't read. For a local business whose pages exist to win customers, blocking is usually self-sabotage.

Is robots.txt the only thing AI crawlers check?

No. They also need your pages to be server-rendered or otherwise readable without heavy JavaScript, fast enough to fetch, and not behind a login. robots.txt is the gate — but it's only the gate.

Originally published at the AI Visibility Index. Free research by the makers of aicantfindme.com — find out why AI search isn't recommending your business.