5 of 9 AI-visibility competitors checked, including Semrush, Otterly.AI and Scrunch AI, serve an llms.txt file, a plain-text index built for markdown-style parsing. 4, including Profound and Ahrefs, don't. Checked directly, live, 29 August 2026.
Disclosure: GetIntel is an AI visibility tool and one of the domains checked below. GetIntel itself serves an llms.txt file. This is a real-time technical check GetIntel ran on each domain's own site, not a claim about whether markdown or HTML gets cited more often, that's a separate, harder question addressed further down.
Which AI Visibility Companies Actually Serve an llms.txt File?
A near-even split. 5 of the 9 AI-visibility competitors checked serve an llms.txt file; 4 don't.
| Domain | llms.txt present? |
|---|---|
| Otterly.AI | Yes |
| LLM Pulse | Yes |
| Dageno | Yes |
| Semrush | Yes |
| Scrunch AI | Yes |
| Profound | No |
| SE Ranking | No |
| Ahrefs | No |
| Peec AI | No |
Checked live by requesting /llms.txt on each domain, 29 August 2026. A "Yes" means the file returned a real 200 response, following redirects where present; a "No" means it 404'd.
That split is worth sitting with. These are companies whose entire product is telling other brands to optimize for how AI engines read their content, and fewer than two-thirds of them have adopted the one concrete artifact most closely associated with that advice on their own site.
Does Having an llms.txt File Actually Mean a Site Gets Cited More?
Not by itself, and no one, including this article, has settled that question yet.
llms.txt is a discovery file, a curated index of a site's important pages, not the content format of the pages themselves. Having one doesn't mean a site's actual articles are served as clean markdown; not having one doesn't mean an engine can't parse the page fine as HTML. The two questions, "does this site have an llms.txt" and "does markdown or HTML get cited more," are related but not the same, and this article only has real data on the first one.
On the second question, the honest state of public evidence is mixed. A GEO experiment run by a competitor in this category, Otterly.AI, tested markdown against HTML directly on its own site and found LLMs process plain text more efficiently. A separate analysis from DeepSmith, an AEO and AI-visibility research site, reached a more skeptical conclusion: every major AI engine reads HTML today, and markdown endpoints cut tokens without a proven citation lift. Neither of those is GetIntel's own measurement, and they don't agree, which is itself the most honest thing to report here.
So Should You Serve Markdown, HTML, or Both?
Both, until the evidence settles, since the cost of doing both is low and the risk of picking wrong is not.
An llms.txt file costs little to add and directly signals to crawlers what matters on a site, independent of whether it changes citation rates. A well-structured HTML page with clear, extractable direct-answer paragraphs remains readable by every engine regardless of whether that engine also fetches a markdown alternative. Betting the whole GEO strategy on one format over the other, before the citation-lift question has real data behind it, is the wrong place to take a strong position.
What Doesn't the llms.txt Adoption Data Prove?
Two real limits apply to this article's llms.txt adoption finding (5 of 9 AI-visibility competitors serve one), separate from the open question about citation lift.
The 9 domains checked are the AI-visibility-tool competitor set GetIntel already tracks, not a representative sample of the wider SaaS market; adoption in a different category could look different. And this check measured presence of the file at one point in time; llms.txt adoption in this category has been moving, so a re-check in a few months would likely show a different split.
How Did GetIntel Check llms.txt Adoption Across These 9 Domains?
Each of the 9 domains was requested directly at /llms.txt, live, on 29 August 2026, following HTTP redirects where the domain issued one. A 200 response counted as present; a 404 counted as absent. This is GetIntel's own real-time check, not a report of another source's findings.
For what actually belongs inside an llms.txt file and how to generate one, see the definitive llms.txt guide. For the broader question of which sources AI engines actually pull citations from, format aside, see which sources AI engines actually cite.