Are you visible to the engines that now mediate search?

What "AI visibility" actually means in 2026

Search is no longer one surface. A user looking for an answer in 2026 may type into Google, ask ChatGPT, query Perplexity, talk to Claude, ping Gemini inside Gmail, or use Copilot inside Word. Each of those engines builds its answer from a different shortlist of sources. Ranking #1 on Google guarantees nothing about whether ChatGPT will mention you to its roughly 700 million weekly users [2].

AI visibility is the discipline of making sure your site is readable, citable, and attributable across all of those engines. It overlaps with classic SEO but it is not the same job. A site can have flawless Core Web Vitals, perfect meta titles, and a 90+ Lighthouse score, and still be invisible to LLMs because it blocks GPTBot in robots.txt, ships no structured data, and writes its content as marketing copy instead of question-shaped answers.

The seven engines we grade you for

• Google organic. Classic ten blue links. Still the largest single source of search demand, though shrinking.
• Google AI Overviews. The generative answer box now appearing on roughly 30% of US informational queries [1]. Cites four to seven sources per answer.
• ChatGPT search. OpenAI's in-product search, which cites web sources for answerable queries.
• Claude. Anthropic's assistant, increasingly used for research and citation lookups.
• Perplexity. Answer engine built around citations from day one. Punches above its weight on referral traffic.
• Gemini. Google's assistant, embedded across Workspace.
• Copilot. Microsoft's assistant, embedded across Bing and Office.

What we check that classic SEO tools do not

Seobility, Ahrefs and Semrush are excellent at what they were built for: the 2015 ten-blue-links Google. None of them grade the AI-search surface. We do. Here is what changes:

1. AI crawler access

We fetch your robots.txt, headers and meta tags and tell you exactly which AI crawlers can read you. The list includes GPTBot, ChatGPT-User, OAI-SearchBot, ClaudeBot, anthropic-ai, PerplexityBot, Google-Extended, GeminiBot and Bingbot. Most sites accidentally block one or more of these, usually because a default robots.txt template was copied from a 2019 blog post [5][4].

2. llms.txt presence and quality

The llms.txt convention [6] gives LLMs a curated map of your site, similar to what sitemap.xml does for traditional crawlers. We check whether you ship one, whether it points to your highest-value pages, and whether the page summaries are written as briefs rather than marketing copy.

3. Answer-extraction readiness

AI Overviews and answer engines extract passages, not pages. We score whether the first 200 words of each page directly answers the page's primary query, whether your H2s are phrased as questions, and whether you provide TL;DR blocks above the 800-pixel fold.

4. Schema for LLM extraction

We crawl your JSON-LD and grade it for the schemas LLMs actually use to build entity links: Organization with sameAs edges to Wikidata, LinkedIn and Crunchbase; Person with credentials, jobTitle and knowsAbout; Article with author and datePublished; FAQPage where appropriate.

{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "Your Company",
  "url": "https://yourcompany.com",
  "logo": "https://yourcompany.com/logo.png",
  "sameAs": [
    "https://www.wikidata.org/wiki/Q…",
    "https://www.linkedin.com/company/your-company",
    "https://www.crunchbase.com/organization/your-company"
  ]
}

5. Entity coverage (GEO)

Generative Engine Optimization. LLMs need named entities to disambiguate you. We check whether your pages name the products, people, technologies, and locations the model needs to attribute information correctly.

6. Citation density

LLMs are statistically more likely to cite pages that themselves cite primary sources. We grade whether your long-form content carries inline references and links to authoritative domains.

What you get back

A graded report across all seven engines, an executive brief, an engineer-ready ticket list, and a 30/60/90 plan. Every finding ships with the raw evidence we used: HTTP headers, JSON-LD snippets, content excerpts, and the URLs where each issue appears. No black-box scores you cannot audit.

How long it takes

Sixty to ninety seconds. We crawl up to 50 of your most important pages, run the checks in parallel, and let an LLM draft the executive summary while we are still finishing the technical checks. You can leave an email and we will send the report when it is done, or watch it stream in live.

The 7-point AI-visibility checklist

1. Allow OAI-SearchBot, ChatGPT-User, ClaudeBot, PerplexityBot in robots.txt.
2. Server-render the answer in HTML — never wait for client-side hydration.
3. Ship Organization + Person JSON-LD with sameAs to Wikidata, LinkedIn, Crunchbase.
4. Open every page with a 2-3 sentence direct answer in the first 200 words.
5. Rewrite at least three H2s as actual user questions.
6. Publish /llms.txt with curated links to your highest-value pages.
7. Cite primary sources inline — LLMs cite pages that cite sources.