What is llms.txt?

Why it matters to a local business

A single-location bakery can write an llms.txt file in a few minutes: its name, a two-sentence summary, and links to its menu and hours pages. Whether that file changes anything measurable is genuinely unclear, so it is worth treating as a low-cost extra rather than a required project, and definitely not a substitute for a crawlable site, accurate schema, and real reviews on the pages an AI engine actually opens.

The exact format, with an example

The llms.txt specification is "a proposal to standardise on using an /llms.txt file to provide information to help agents use a website" (llmstxt.org, 2026), created by developer Jeremy Howard and first published in September 2024. The only required section is "an H1 with the name of the project or site," followed optionally by "a blockquote with a short summary" and further H2-headed sections linking to key pages with short notes (llmstxt.org, 2026). A minimal, correctly formatted file for a single-location business looks like this:

# Riverside Family Dental

> Riverside Family Dental is a general and cosmetic dentistry practice in Springfield, Ohio, open six days a week with same-day emergency appointments.

## Key pages
- Services: general, cosmetic, and emergency dental care
- Hours and location: address, phone number, and current hours
- Book an appointment: online scheduling

The stated goal behind this structure is that "agents are best served by concise, expert-level information gathered in a single, accessible location" (llmstxt.org, 2026), addressing the mismatch between pages built for human visitors, full of navigation and ads, and what a language model actually needs to work efficiently.

Does it actually change what AI engines show

Here is the part most llms.txt guides will not tell you plainly. No major AI lab has published a commitment to actively parsing and ranking based on llms.txt files. Google's own AI features documentation states directly that "you don't need to create new machine readable files, AI text files, or markup" to appear in AI Overviews or AI Mode (Google, 2026), a statement general enough to cover llms.txt specifically, even without naming it. The honest summary: publishing one is not confirmed to change whether any current AI engine cites you, though it costs little enough that the uncertainty is not a strong reason to skip it either.

Common errors and troubleshooting

The most common problem is a file that simply is not where it needs to be. A checker or a crawler looking for llms.txt expects it at your domain's root, such as yoursite.com/llms.txt, not inside a subfolder, and not saved with an extra extension like llms.txt.md or llms.txt.html. If a tool reports the file as not found, first open that exact URL directly in a browser: a correctly published file loads as plain text immediately, while a missing one returns a 404 error page. Other common causes include a robots.txt rule accidentally blocking access to the file, a caching layer serving an old version or a 404 from before the file existed, or the file being uploaded to a staging environment instead of the live production site.

It is also worth checking that whatever crawler you care about can reach your site at all, since llms.txt does nothing for an engine that is blocked before it ever gets there. See AI crawlers for the specific bots and robots.txt rules that control that separate, more consequential layer.

What to do about it

• Build yours in minutes with the free llms.txt Generator, which turns your business summary and key pages into the correct format.
• Keep it short: your business name as the H1, a two-sentence blockquote summary, and links to your services, hours, and contact pages.
• Place the finished file at yoursite.com/llms.txt, the predictable root location every spec-following reader expects, and confirm it loads by visiting that URL directly.
• Do not treat a finished llms.txt file as a completed AI visibility project; it is a small addition, not a strategy.
• Confirm your crawler access and LocalBusiness schema are solid first, since those carry more confirmed weight for AI visibility today.