GEO Checker
What AI assistants find when they look at openai.com.
Pages read
This site turned our scanner away by name (HTTP 403), so the report reads it as a browser does. What the assistants’ own fetchers get from the home page: Claude-User · ChatGPT-User · meta-externalagent
Cached 6 days ago
GEO Score
90
- Site readiness
- 76 Twelve checks on every page read.
- Recognition
- Well known Assistants already know OpenAI, and these pages give them material to quote as well.
- Shopping agents
- Not a shop.
Who assistants think this is
OpenAI
- Does
- Artificial intelligence research and deployment
- Based in
- San Francisco, United States
- What the assistant knew beforehand
- OpenAI is a leading AI research company founded in 2015, known for developing GPT language models and ChatGPT, a widely-used conversational AI assistant.
- Memory
- Recognised from the domain alone — AI research company; develops large language models.
- Wikipedia
- 98 language editions.
- Maps
- No Google Maps listing is linked to this website.
What they can read
- Can an assistant get the page at all? robots.txt for the AI search and live-fetch crawlers, noindex, the canonical, the sitemap, and whether the content exists before JavaScript runs — the crawlers of OpenAI, Anthropic and Perplexity do not run it. A gate, not a merit: scored as its weakest check, and left out of the score when nothing is wrong. 90
- Can it parse what it got? Structured data (an Organization plus a type that says what the page is), one H1 and H2 sections, question headings with an answer beneath, landmarks, title, description and Open Graph tags. 59
- Is there anything worth quoting? Specific claims — years, prices, quantities, places — rather than adjectives any company could use; on an article, a byline, a date and links to sources. Plus a language model’s read of every page: what it could cite, and what the site does not state. 72
- Can it tell who is behind the site? Organization markup with sameAs links to verified profiles and a logo, an About link, a way to reach a person. On-page signals only — reputation is Recognition’s question. 77
- Is it current? The newest date among Last-Modified, <time>, Open Graph and schema dates. Event dates in the future do not count; an undated brand page is fine, an undated article is not. 82
- home / 81
- about /about 72
- services /solutions 80
- prices /business/pricing 80
- contact /contact-sales 42
The AI read
OpenAI is an AI research and deployment company whose home page functions as a news hub, surfacing recent product releases, research publications, and customer stories. The site names specific models (GPT-6, GPT-5.6 Sol, GPT-5.5) and states the company's mission to ensure artificial general intelligence benefits all humanity. The About page describes OpenAI's structure as a nonprofit foundation governing a for-profit public benefit corporation. The Solutions page lists six departments and six industries where ChatGPT is deployed, with case studies from Virgin Atlantic, Zapier, and others, but the main pages read do not state pricing, deployment timelines, or technical specifications; those details likely sit behind the Contact Sales and Try ChatGPT Business links.
OpenAI's site surfaces product names, research topics, and customer case studies, but states little about pricing, deployment specifics, or concrete capabilities.
The Solutions page could state concrete outcomes—response times, accuracy rates, or cost savings—from the named customer case studies (Virgin Atlantic, Zapier, Model ML) rather than linking to them.
What an assistant receives
home
- Content
- 262 words 51% of the words on the page. The other 253 are menus, header and footer.
- Title and description
- OpenAI | Research & Deployment We believe our research will eventually lead to artificial general intelligence, a system that can solve human-level problems.
- Structured data
- None on this page.
- Quotable
- 1 passage the model would quote, underlined.
Each page as our scanner fetched it, without running JavaScript — which is how the crawlers of OpenAI, Anthropic and Perplexity read. H1–H6 are headings, Q and A a question with the answer under it. A site that serves a named assistant something different is not visible from here.
Fix first · 3
The paths under a finding are the pages below the pass line — fix the template once and every page follows.
Working · 3
- Meta gaps: title length 22 (target 25–70); meta description length 185 (target 70–170).
- Organization schema, sameAs links, About, and contact signals present; no tel: link.
- No date signals on this brand page — optional here.
Access
Whether assistants can get the pages at all. Hygiene, not merit: a site either lets them in or it does not, so these never pad the score — they only cost.
- AI crawlers allowed good
- All AI search crawlers can reach this page.
- Indexable good
- Page is indexable.
- Readable without JavaScript needs work
- Raw HTML carries 85 words — thinner than ideal for non-JS crawlers.
- Canonical good
- Canonical tag present and self-referencing.
- Sitemap good
- Sitemap declared in robots.txt.
- Largest paint needs work
- 2.9 s
- Responsiveness needs work
- 234 ms
- Layout shift good
- 0.00
Questions people ask AI assistants to find sites like this
Derived from the pages read: what a customer types when looking for what you offer, without knowing you exist. Whether you appear in the answers — and who does instead — is what the full check measures.
- Which AI platform should I use to automate customer service for a retail business?
- What AI tools do financial services companies use for compliance and risk analysis?
- Where can I find an enterprise AI solution with built-in security and data privacy controls?
Do you appear in the answers?
The full check reports whether you are named, who is named alongside or instead, and which sources put them there — with a plan for improvements.
hello@dariodario.com