Are these companies your clients?

No. These are public observations of their websites. We have no relationship with them and make no claims about their results.

How many of the 10 sites publish llms.txt?

All 10 returned an llms.txt starting with a Markdown H1 when we fetched them on 15 November 2026.

Do these SaaS sites block AI crawlers?

None blocked its whole site for all crawlers. Only Zapier named AI crawlers, and its rules close only the /app/ area.

Which schema types did their home pages use?

Mostly Organization plus WebSite, SoftwareApplication, Product or WebPage. Linear's home page HTML had no JSON-LD in our fetch.

Can I repeat this check?

Yes. Fetch /robots.txt, /llms.txt and the home page HTML, then look for user-agent groups and JSON-LD @type values. Results change over time.

Blog

B2B SaaS GEO: what 10 well-known SaaS sites publish for AI crawlers (public observations)

Jisung Yoon · Yunju Company · 15 November 2026

Short answer: on 15 November 2026 we fetched public files from 10 well-known B2B SaaS sites. All 10 publish an llms.txt that starts with a Markdown H1. None blocks the whole site for all crawlers in robots.txt. Only one (Zapier) names AI crawlers such as GPTBot, OAI-SearchBot and PerplexityBot in robots.txt, and even there the rules close only the /app/ area. Most send structured data on the home page, usually Organization plus WebSite or SoftwareApplication. Several also publish documentation in Markdown.

Important: these are public observations of other companies' websites. They are not our clients, we have no relationship with them, and we make no claims about any outcome (traffic, citations or rankings) for any of them. The files can change at any time; what follows is what we fetched on the date shown.

Why this analysis

GEO buyers often ask what established software companies do. Rather than repeat opinions, we fetched the files ourselves. The method is below so you can repeat it.

Method

Date: 15 November 2026, about 20:59–21:01 KST.
How: plain HTTPS requests (curl with a desktop browser user agent) to /robots.txt, /llms.txt and the home page of each site. We parsed robots.txt for user-agent groups and read JSON-LD @type values from the home page HTML. Pages that build schema with JavaScript after load would not be seen by this method.
Sites: stripe.com, vercel.com, supabase.com, linear.app, hubspot.com, zapier.com, notion.com, intercom.com, n8n.io, make.com.

What we found

Site	llms.txt (first line)	AI crawlers named in robots.txt	Home page JSON-LD types
stripe.com	`# Stripe`	none	Organization, WebSite
vercel.com	`# Vercel`	none	Organization, Service, SoftwareApplication
supabase.com	`# Supabase Docs`	none (has a `Content-Signal` line)	Organization, Service, SoftwareApplication, WebSite
linear.app	`# Linear`	none	none found in the HTML
hubspot.com	`# HubSpot`	none	Organization, Product, WebPage, WebSite
zapier.com	`# Zapier`	GPTBot, OAI-SearchBot, ChatGPT-User, PerplexityBot, ClaudeBot, Claude-SearchBot, Google-Extended (each: `disallow: /app/`, `allow: /app/login`); plus a `Content-Signal` line	Organization, SoftwareApplication, WebPage, WebSite
notion.com	`# Notion`	`meta-externalagent` (disallow: /)	Organization, SoftwareApplication, WebSite
intercom.com	`# intercom.com — llms.txt`	none (has a `Content-Signal` line)	BreadcrumbList, ItemList, Organization, Product, SoftwareApplication, WebPage, WebSite
n8n.io	`# n8n`	none	Organization
make.com	`# Make`	none	Organization, WebPage, WebSite

"None" means robots.txt has no group naming those crawlers, so they fall under the general User-agent: * rules. On all 10 sites those rules don't disallow the whole site.

Docs in Markdown

We also checked whether documentation is offered as Markdown, which the llms.txt proposal recommends for pages agents might need:

docs.stripe.com/llms.txt and docs.stripe.com/payments.md: both served as text/markdown.
docs.n8n.io/llms.txt and docs.n8n.io/workflows/export-import.md: both served as text/markdown. The n8n docs page itself points readers to llms.txt and to .md versions.
vercel.com/docs/llms.txt and developers.hubspot.com/docs/llms.txt: served (plain text, Markdown content).
linear.app/docs/llms.txt: 404. Linear's root /llms.txt exists.

How the llms.txt files are organised

The H2 sections differ by company. Vercel's include "When to use Vercel" and "How agents should use Vercel". Supabase's include "Documentation", "Pricing" and "Product overviews". Intercom's include "Customer Evidence" and "Comparisons". Zapier's start with "What is Zapier". Most group links by product or audience rather than listing every page.

Patterns worth copying (and what not to read into them)

A published llms.txt is common among these sites. It costs little, but remember that Google says AI text files aren't needed for its AI features (how to write one).
Not blocking AI search crawlers is the norm here. Nine of 10 don't mention them at all. The one that does closes only its logged-in app area. Compare that with a blanket "block AI bots" firewall setting.
Organization schema is nearly universal on these home pages. Google says structured data should match the visible page.
Docs as Markdown is a developer-tools pattern (Stripe, n8n). It's less relevant for a local service business.

We don't know how any of this affects these companies' visibility in AI answers, and we don't claim it does.

Repeat it on your own site

Run the AI search visibility checklist, or order a GEO audit ($149, 48 hours, pricing). Our own site's changes are documented in the case study.

Sources

Fetched files (2026-11-15, ~20:59–21:01 KST): /robots.txt, /llms.txt and the home page of stripe.com, vercel.com, supabase.com, linear.app, hubspot.com, zapier.com, notion.com, intercom.com, n8n.io and make.com; plus docs.stripe.com, docs.n8n.io, vercel.com/docs, developers.hubspot.com/docs and linear.app/docs llms.txt checks.
llms.txt proposal: https://llmstxt.org/ (checked 2026-11-15)
Google Search Central, "AI features and your website": https://developers.google.com/search/docs/appearance/ai-features (last updated 2025-12-10; checked 2026-11-15)