
Prior to installing one other plugin or pay somebody to create an llms.txt file, ask what you count on it to do.
If the reply is “get my web site into Google’s AI solutions,” you’ll be able to cease. Google explicitly says it ignores these recordsdata for search visibility and rankings, together with its generative AI options. An llms.txt generator can’t change that. Google’s AI optimization information
There are reputable causes to publish the file, particularly when individuals use AI brokers to work together with your software program documentation. There are additionally handy methods to generate it.
However a web site proprietor on the lookout for extra prospects deserves a clearer clarification than “AI wants this.”
I appeared on the guarantees on a generator’s web site, the proposal behind the file, Google’s documentation and analysis into whether or not something truly requests these recordsdata. My recommendation: set up a use for the output earlier than selecting a software to provide it.
First, it is a web site file
The proper identify is llms.txt. The “llms” refers to massive language fashions; the extension is .txt. An llms.txt generator creates a file about your web site.
The proposal describes a Markdown doc containing a web site or venture identify, a abstract and arranged hyperlinks to helpful sources. Markdown is abnormal textual content with easy formatting: a # introduces a heading, for instance.
Think about a software program firm with a whole lot of documentation pages. A brief file might level an agent towards the set up information, authentication directions and API reference.
That’s an comprehensible use. It provides software program someplace helpful to begin when somebody directs it to the documentation.
It doesn’t, by itself, clarify why an AI search service ought to suggest that firm to a possible buyer.
Learn what the generator is promising
The LLMrefs generator pages normally describe free instruments that crawl a web site, makes use of AI to jot down web page descriptions and produces an editable file. These are concrete options: it could possibly prevent some copying, organizing and writing.
The identical web page goes additional. It calls the file “important for GEO/AEO” and tells customers that “AI instruments will mechanically discover and use this file.” It additionally presents higher suggestions and a aggressive benefit as advantages.
These claims deserve separate scrutiny.
Producing a file is one thing the generator controls. Whether or not an outdoor service discovers it, makes use of its contents and modifications its suggestions is outdoors the generator’s management.
Google has already answered the principle search engine optimisation query
Google’s AI optimization information particularly addresses llms.txt beneath its mythbusting recommendation.
It says sustaining these recordsdata will “neither hurt nor assist your web site’s visibility or rankings” in Google Search.
Google additionally explains that it might uncover and crawl a file with out giving the format particular therapy. Seeing Googlebot request your llms.txt due to this fact wouldn’t contradict its steerage.
For a enterprise hoping to seem in AI Overviews or AI Mode, this removes the principle cause to deal with an llms.txt generator as an search engine optimisation requirement.
That assertion applies to Google Search. It shouldn’t be stretched right into a declare that each AI product ignores the file.
The most important quantity wants its small print
Ahrefs studied 137,210 domains utilizing its Internet Analytics product and examined requests to /llms.txt throughout Might 2026.
Round 38,000 domains had a file that handed its checks. Of these recordsdata, 97% obtained no requests within the measured interval.
That may be a severe problem to the concept that publishing one mechanically attracts AI readers.
Nevertheless, it was a research of an Ahrefs buyer inhabitants over one month. It measured requests, not a managed change in rankings or citations. It additionally predates the proposal’s August 2026 revision.
The discovering helps skepticism about automated discovery. It can’t set up that each file is ineffective, or that no agent will ever use one.
The excellence additionally works within the different course: a recorded request doesn’t show the contents influenced a solution. Downloading a file, following its hyperlinks and recommending a enterprise are totally different outcomes.
There’s a actual use case, and it might already be lined
Suppose your prospects use coding brokers to combine your product. Serving to these brokers find the right documentation might be helpful.
You may see an precise implementation in Anthropic’s developer documentation index, which lists documentation sources.
Publishing such an index demonstrates that an organization gives the file. It doesn’t display that its AI merchandise routinely use different web sites’ recordsdata to pick search suggestions.
Documentation platform Mintlify automatically generates and hosts llms.txt, retains the generated file present and advertises it via response headers. In case your documentation already makes use of that characteristic, you could have no cause to discover a separate generator.
There has additionally been a significant replace to the proposal. Its August 2026 revision provides steerage for linking a web page to its Markdown model and the related llms.txt file. The change addresses discoverability: how an agent can discover the sources with out guessing.
That makes checking the implementation extra helpful than checking a field marked “file exists.”
For a documentation workforce, the sensible questions are which brokers prospects use, how these brokers discover your sources, and whether or not they can full the meant job precisely.
Should you really want a file, you’ll be able to write one
For a small documentation web site, a textual content editor is sufficient. Right here is an illustrative starter:
# Instance Reserving API
> Documentation for builders integrating the Instance Reserving API.
## Begin right here
- [Setup](https://instance.com/docs/setup.md): Create a take a look at integration.
- [Authentication](https://instance.com/docs/authentication.md): Authenticate requests.
## Reference
- [Reservations](https://instance.com/docs/reservations.md): Create and replace bookings.
- [Errors](https://instance.com/docs/errors.md): Interpret error responses.The title, abstract and grouped hyperlinks observe the fundamental construction described within the llms.txt specification.
Substitute the instance URLs and descriptions with your personal. Solely hyperlink to Markdown variations that truly exist; including .md to a URL doesn’t create a web page.
Save the doc as llms.txt and have your web site serve it on the meant public location, equivalent to /llms.txt or /docs/llms.txt. Verify that the URL returns the precise textual content and that its hyperlinks work. Comply with the proposal’s discovery steerage in case you are implementing it for appropriate brokers.
A generator turns into helpful when it saves significant work sustaining a bigger assortment. For a small file, you’ll be able to choose whether or not choosing and describing a couple of hyperlinks your self is faster than establishing one other software.
Both approach, somebody must test the consequence.
#llms.txt #generator #dont
