AI Visibility
The llms.txt census: how many SaaS sites actually serve one
We checked 170 domains for llms.txt on 26 August 2026. Here are the adoption counts, file-quality checks and limits of what a one-day census can establish.
Keep the question with the answer.
- QuestionKeep the wording
- AnswerSave the response
- RecordEngine and date
- RepeatCompare like with like
In this article
An llms.txt file offers an agent a short description of a site and links to useful pages1. We checked how many domains in our research sample served one, and what the files contained. The census measures publication, rather than whether an engine used the files.
What we checked
On 26 August 2026 we requested /llms.txt once from each of 170 distinct domains. The sample combined a 152-company cohort with 19 well-known SaaS and devtools domains checked in July. One domain appeared in both groups. The cohort contained 118 young YC companies from 2021 to 2026 batches, 18 established controls and 16 AI-search vendors.
A file counts as present when the request returns a 2xx with a body that is neither empty nor HTML. The HTML test matters more than it sounds. A single-page app that answers every path with its index shell would otherwise appear to serve a file at every address we asked for. Any other HTTP status counts as missing.
How many serve one
| Population | Domains | Serve llms.txt |
|---|---|---|
| All domains checked | 170 | 45.9% |
| Well-known SaaS and devtools companies | 19 | 84.2% |
| Companies selling AI-search tools | 16 | 68.8% |
| Established controls | 18 | 38.9% |
| Young companies (YC 2021 to 2026) | 118 | 38.1% |
The well-known group had the highest observed adoption: 16 of 19 domains, or 84.2%. That was one more file than in July. The AI-search vendor group was next at 68.8%; young companies and established controls were both near 38%. These groups differ in more than age.
Within the young cohort, adoption was 43.8% for developer infrastructure, 30.0% for business software and 25.0% for healthcare. The difference is consistent with greater adoption in developer-facing companies, but the census does not explain its cause.
The files are not stubs
Before the run, we predicted that most files would be thin. The recorded definition was under 500 bytes or no prose sentence anywhere in the file. Keeping that definition fixed allowed us to compare the prediction with the result.
Of the 78 files we found, one was thin.
| Slice | Files | Opens with a heading | Carries a summary | Median size |
|---|---|---|---|---|
| All files found | 78 | 94.9% | 85.9% | 7,628 bytes |
| Young companies (YC 2021 to 2026) | 45 | 95.6% | 84.4% | 6,506 bytes |
| Established controls | 7 | 85.7% | 71.4% | 13,258 bytes |
| Well-known SaaS and devtools companies | 16 | 93.8% | 93.8% | 6,435 bytes |
The median file contained 26 links. About 86% included a blockquote summary, and every present file passed the prose-sentence check. Sizes ranged from 387 bytes to 1.5 MB. Those checks establish content and structure; they do not establish accuracy, usefulness or ongoing maintenance.
Serving it is not the same as being read
Presence is only one stage of use. A request returning a file does not establish whether an assistant discovered it, opened it or relied on anything it contained. This census did not instrument those later stages.
If you publish llms.txt, give visitors a clear link to it and keep the referenced pages useful and accessible. You can test whether your own agent follows that route. Such a test describes the observed session, rather than every crawler’s behaviour.
What we got wrong
Two recorded predictions were contradicted by the results: adoption exceeded the expected 30% threshold, and only one of 78 present files met the thin-file definition. Developer-facing companies had higher adoption than the other measured sectors, matching the predicted direction.
What this does and does not say
- It measures adoption and file quality on one day. It is not evidence that serving llms.txt causes AI visibility, and no figure here should be read that way.
- The groups differ in age, sector and other characteristics. This census does not isolate which characteristics explain adoption or whether any also affect AI visibility.
- Presence means a 2xx, non-empty, non-HTML response at /llms.txt. It does not mean an engine fetched the file, read it, or acted on it.
- Two domains answered 403, so the true count could be higher by at most two. One domain was unreachable and is excluded from the reachable-share figures.
- The population is a fixed research cohort plus 19 well-known companies. It is not a random sample of software companies, and nothing here claims to generalise beyond it.
The checker fetches your llms.txt the way a crawler would and tells you what came back. If you do not have one, it generates a starter file you can publish. Free, no signup.
Run the llms.txt checkerRead next
How long before Google shows a change
Typical and slow times for Google to find, index and show a change, as reported from a Google talk, and when to judge a growth move by them.
Read →JournalI launched my SaaS and nobody is finding it. What now?
A quiet launch is the normal start for a young product. What our AI search data shows, the three checks to run in order, and the one move to make after them.
Read →