Does llms.txt Do Anything? What Google Actually Said

For the past two years, llms.txt has been sold to marketing teams as the new must-have file, described as the robots.txt for AI, the thing your site supposedly needs so ChatGPT and Google's AI Overviews can find you. Agencies added it to technical SEO checklists. Generators popped up to auto-create one in seconds. And somewhere in that noise, a simple question got lost: does llms.txt seo actually move the needle for search visibility?

Google has now answered that question directly, in writing. This piece lays out exactly what llms.txt proposes to do, what Google has said about it on the record, who actually reads the file today, and what that means for your technical SEO priorities in 2026.

What llms.txt Proposes to Do

llms.txt is a plain text, Markdown formatted file placed at a site's root, found at yourdomain.com/llms.txt, that gives large language models a curated map of a site's most important content. It was proposed in September 2024 by Jeremy Howard of Answer.AI, not by a standards body like the W3C or IETF. It is a community proposal, hosted openly, with no formal ratification behind it.

The format is deliberately minimal. Under the specification, the only required element is an H1 with the site or project name. Everything else, including a short blockquote summary, optional context, and H2 level sections listing key pages as Markdown links, is optional. A companion file, llms full.txt, is sometimes used to bundle the full text of referenced pages into a single fetchable document, though that variant isn't part of the core spec.

The logic behind it is reasonable on its face. Language models work with finite context windows, and most websites are built for human navigation, including menus, JavaScript, and ads, rather than for token-efficient reading. llms.txt was pitched as a hand-curated shortcut that saves an AI system from having to parse an entire site to understand what it's about.

That reasoning is sound in the abstract. The problem is what happened next: llms.txt got marketed as an SEO ranking factor, when it was never designed, published, or endorsed as one.

What Google Has Actually Stated

This is the part that matters most, and Google has not been vague about it.

In an AI optimization guide update published on June 15, 2026, Google added a section addressing llms.txt by name. The core statement is that publishers do not need machine-readable files, AI text files, or Markdown versions of their content to appear in Google Search or its generative features, because Search itself doesn't read them. Google also clarified that keeping an llms.txt file for other tools that do use it is harmless. It simply won't help or hurt rankings or visibility in Google Search, since Search ignores the file entirely.

This wasn't a new policy or a reversal. Google has never used llms.txt as a ranking input, so nothing actually changed on June 15. The update was Google formally correcting an industry narrative that had gotten ahead of the facts. The company framed it as a response to recurring questions from site owners and SEOs.

It also isn't the first time Google addressed this. Gary Illyes, from Google's Search Relations team, said as early as mid 2025 at a Search Central Live event that Google was not supporting llms.txt and had no plans to. John Mueller, also on that team, went further, comparing llms.txt to the old meta keywords tag, a self-declared, easily gamed signal Google stopped trusting decades ago. Mueller also addressed the confusion caused by some Google-owned properties serving an llms.txt file. That happened because an internal content system added support for it, not because Google's Search systems were consuming it.

The same guide extends the point to Google's generative features. AI Overviews and AI Mode are built on the same underlying Search index and ranking fundamentals as classic results. There is no separate llms txt google fed pipeline powering those surfaces.

Which Consumers, If Any, Use It Today

Google's position covers Google. It doesn't settle the question for every AI system, and this is where a fair account needs nuance instead of a flat useless verdict.

Independent adoption research paints a mixed picture. A large-scale study covering hundreds of thousands of domains found that the overwhelming majority of published llms.txt files were never fetched by anything, and statistical modeling found no measurable citation effect from having one. That's a meaningful data point against treating the file as a visibility lever for AI answer engines generally.

At the same time, the file isn't fetched by nobody. Coding agents and developer tools, things like AI-assisted IDEs and agentic coding assistants, are the most consistent real-world consumers, since they benefit from a clean, low-noise map of documentation and API references. Some AI answer engines have not ruled out reading the file during retrieval, though none have published the kind of clear commitment Google gave in the opposite direction. Chrome's Lighthouse tooling now also audits for the file's presence, a separate initiative from Google Search, aimed at browser agent readiness rather than ranking. That split is part of why the guidance has looked contradictory to people skimming headlines.

The accurate summary: llms.txt is not a Google Search signal, not proven to influence AI answer engine citations at scale, but does have a narrow, real audience in developer and agent tooling.

Why It Is Being Oversold

The gap between what llms.txt is and what it's marketed as comes down to incentive, not evidence. AI SEO is a fast-moving, poorly defined space, and a simple checklist item, add one file and claim you're AI ready, is an easy thing to sell. Generators, plugins, and audit tools built entire product lines around a file that no major search engine has confirmed using for ranking.

There's also a real structural confusion baked into the rollout. Google Search ignoring the file while Chrome's Lighthouse audits for it are two different product teams answering two different questions, but that split reads to a lot of site owners as Google flip-flopping. It isn't. One team is about ranking. The other is about whether a browser agent can navigate your site. Conflating them is where most of the misinformation comes from.

Should You Add One Anyway?

Given all that, an llms.txt file is not worth chasing as a ranking tactic, but it's also not something you need to rip out or avoid. The honest framing is closer to low cost, low risk, narrow benefit than either essential or useless.

Reasons it can still be worth the ten minutes it takes to create:

• It costs almost nothing to build and maintain, especially for a documentation-heavy or developer-facing site.

• It can genuinely help coding agents and AI-assisted dev tools navigate your docs more efficiently, which is a real, if narrow, use case.

• It causes no downside for Google Search, since Search ignores it entirely. There's no risk of it hurting rankings.

Reasons not to prioritize it:

• It will not move your Google Search rankings, AI Overviews presence, or AI Mode visibility. Google has said this plainly.

• Independent data suggests it has little to no measurable effect on AI answer engine citations either.

• Time spent hand-curating and maintaining it is time not spent on things with proven returns, such as crawlability, structured content, and genuine topical authority.

What to Do Instead

If the goal is actual AI and search visibility, the fundamentals haven't changed just because a new file format exists. Google's guidance is consistent on this: rankings and AI Overview inclusion come from the same things they always have, including content that's genuinely useful, pages that are easy to crawl and parse, clear on-page structure with real headings rather than just visual formatting, and topical depth that a model or a human reader can actually extract value from.

For teams weighing where to spend technical SEO time in 2026, that means prioritizing clean HTML and semantic markup over speculative AI-facing files, treating structured data and schema markup as the higher leverage lever, and auditing actual crawl and indexation health before layering on anything experimental. llms.txt can sit quietly in the backlog as a low-effort nice-to-have for developer tooling. It shouldn't sit anywhere near the top of an AI visibility strategy.

Key Takeaways

• It costs almost nothing to build and maintain, especially for a documentation-heavy or developer-facing site.

• It can genuinely help coding agents and AI-assisted dev tools navigate your docs more efficiently, which is a real, if narrow, use case.

• It causes no downside for Google Search, since Search ignores it entirely. There's no risk of it hurting rankings.

• It will not move your Google Search rankings, AI Overviews presence, or AI Mode visibility. Google has said this plainly.

• Independent data suggests it has little to no measurable effect on AI answer engine citations either.