Should You Add a “llms.txt” File to Your Site?

If you’ve been following digital marketing chatter lately, you’ve probably run into llms.txt. It’s being talked up as the next step in GEO—a file that tells large language models like ChatGPT and Perplexity what pages they should or shouldn’t use from your site.

For businesses eager to stay ahead in the GEO arms race, the obvious question is: Should we create an llms.txt file?

Short answer: not so fast. Right now, llms.txt is much more hype than real impact. 

Here’s why—and what you should be doing.

What is llms.txt, anyway?

The concept is simple: a file like robots.txt, but aimed squarely at large language models. Placed at your site root (example.com/llms.txt), it might look like this:

User-agent:*

Disallow:*

or                     

User-agent: GPT-bot

Allow:/

In theory, LLM providers (OpenAI, Anthropic, others) would then check this file before scraping or training on your content. Then, hopefully, they would do what you want.

Why All the Buzz?

It’s easy to see the appeal. As LLMs become more central to search and content generation, publishers want control over whether their content shows up in AI answers or training sets.

On paper, llms.txt looks like a clean, transparent way to manage LLM access.

But just because it makes sense in theory doesn’t mean it works in practice.

Here’s the thing: It’s just a proposal

As of September 2025, there is no universal adoption of llms.txt by AI companies. In fact, as recently as July, Google has stated no AI system is even using them. 

Sure, some AI companies have said they would consider supporting the concept in theory, but as of now, there is no benefit at all to developing your own llms.txt file. That means llms.txt:

> Doesn’t improve discoverability for LLM-generated content 

> Doesn’t help you gain additional LLM citations or referral traffic 

> Doesn’t drive performance…or do anything right now

In short, llms.txt is not an SEO or GEO boost—it’s a speculative idea.

Worse, it could create confusion. Conflicting rules between llms.txt and robots.txt might cause issues if LLMs ever adopt the standard. 

And perhaps most importantly: relying on llms.txt could give businesses a false sense of security, distracting them from focusing on tools that actually matter.

What works right now: robots.txt & noindex

Looking for the proposed benefits of an llms.txt file? Here’s where you should focus.

• Robots.txt: It’s the long-standing, widely respected protocol for crawlers. Search engines honor it, and because LLMs use search engine results, this already governs most LLM behavior.

• Meta robots noindex / X-Robots-Tag: These are other tools that provide even better, more granular control over indexing on a page-by-page basis.

In other words, LLMs already use robots.txt, at least indirectly. Your robots.txt and noindex rules are being registered by the sources LLMs rely on. 

As for the future, it’s my opinion that robots.txt is most likely to continue filling both its current role and the proposed role of llms.txt. It’s already a widely recognized file in a standardized location. There’s plenty more that robots.txt can do. It just makes more sense.

Who should care about llms.txt?

Very few sites, honestly:

> Publishers who actively want to opt out of LLM usage entirely (right now or down the road).

> Sites with sensitive or proprietary content that may not want AI models to use or quote their materials.

> Organizations planning for long-term governance, watching standards as they emerge.

Even then, llms.txt should only be a back-up or speculative layer—not your main line of defense.

llms.txt: The final verdict (for now)

Sure, llms.txt sounds interesting, especially since it signals a new, mainstream awareness of the shifting SEO/GEO landscape. But it’s not transformative. It doesn’t deliver traffic. It doesn’t improve visibility. It doesn’t really do anything. It’s aspirational. 

Unless things change for llms.txt (we’ll keep you posted if they do)—it’s not worth the effort. There’s a great deal of change happening daily in AI and Search, but there’s an equal amount of noise. With any new tool or proposal, it’s important to check the data first and see what is actually driving performance, before investing time and resources into projects that fizzle with tomorrow’s headlines. 

Driving LLM Referral Traffic: A Real Plan for the Future

The best advice for businesses right now: focus on your content, your SEO/GEO fundamentals, your user experience—and not llms.txt. You should be optimizing content marketing for LLMs and providing expert, unique, and compelling perspectives. You should be running keyword research for queries that drive users to click, not just to skim an AI Overview. You should be aggressive in staying a step ahead in the information marketplace by investing in strategies that work.

llms.txt is not something most sites should prioritize. It’s a “maybe, later” tool, not a “use now” one. Let’s keep building what's proven. Stay tuned to see real gamechangers in what’s coming next.

Interested in learning more about Saltbox’s SEO/GEO solutions?

Contact our team today.