TruePrime Go

The Business Owner's Guide to llms.txt — Making Your Website AI-Readable

By TruePrime AI · Updated 2026-08-19

You know about robots.txt — the file that tells search engines what to crawl. There's now an equivalent for AI assistants: llms.txt. It's a simple text file at your website's root that tells AI models what your business is, what you offer, and where to find your most important content.

Most businesses don't have one yet. That's an opportunity.

What llms.txt actually is

The llms.txt specification (proposed by Jeremy Howard of fast.ai) defines a standard way for websites to provide information about themselves in a format that large language models can easily read. Think of it as your business card for AI assistants.

Unlike HTML pages — which are designed for humans and contain navigation, styling, images, and interactive elements that AI models have to parse around — llms.txt is pure, structured text. No clutter. Just the facts AI assistants need to understand and recommend your business.

FileAudiencePurposeFormat
robots.txtSearch engine crawlers (Googlebot, Bingbot)What to crawl and indexDirectives (allow/disallow paths)
sitemap.xmlSearch engine crawlersComplete list of pages for indexingXML with URLs and metadata
llms.txtAI assistants (ChatGPT, Perplexity, Claude, Gemini, Copilot)What this business is and what content matters mostMarkdown-formatted text
brand-facts.jsonAI assistants + structured data consumersVerified business facts (pricing, services, contact)JSON with factual claims

Why this matters for your business right now

AI assistants are becoming a primary way people discover and evaluate businesses. When someone asks ChatGPT "what's a good AI marketing service for small businesses" or Perplexity "find me a dentist in Portland that takes new patients," the AI pulls from its training data and any sources it can access in real time.

If your website has a clear, structured llms.txt file, AI assistants can quickly understand:

Without this file, AI assistants have to guess based on whatever they can scrape from your website — which might be your cookie policy, your footer navigation, or a blog post from 2019.

Who should and shouldn't create an llms.txt file

Business typePriorityWhy
Professional services (law, accounting, consulting, financial advisory)HighAI assistants are already answering "find me a [specialty] [professional] in [city]" queries. Your llms.txt is how you get recommended.
SaaS and technology companiesHighComparison queries ("X vs Y") are among the most common AI assistant use cases. llms.txt helps AI assistants accurately represent your product.
Local service businesses (dental, home services, real estate)HighLocal recommendations are one of the most-used AI assistant features. A structured llms.txt makes your business citable for "best [service] near me" queries.
E-commerce with unique productsMediumHelps AI assistants recommend your products for specific needs. Less impact for commodity products where price comparison dominates.
Content publishers and mediaMediumAI assistants already index your articles through web search. llms.txt can help prioritize which content is most authoritative.
Businesses with no web presence goalLowIf you don't want AI assistants to surface your business (rare but valid), you don't need this file.

What goes in an llms.txt file

The format is simple Markdown. Here's what each section does and why it matters:

SectionPurposeWhat to includeCommon mistakes
Title (H1)Identifies your businessYour company name and a one-line description of what you doIncluding taglines or marketing language instead of factual description
Description paragraphCore value proposition for AI context2–3 sentences explaining what your business does, who it's for, and what makes it different. Factual, not promotional.Writing advertising copy. AI assistants extract facts, not persuasion. "The world's leading" means nothing to an AI.
Key pages sectionDirects AI to your most important contentLinks to 5–15 pages that best represent your expertise: guides, case studies, comparisons, pricingLinking to every page on your site. The point is curation, not completeness — that's what sitemap.xml is for.
Services / ProductsStructured list of what you offerClear, factual descriptions of each service or product with pricing if publicVague descriptions like "customized solutions" that give AI assistants nothing specific to cite

A real-world example

Here's a simplified version of what a local dental practice's llms.txt might look like:

# Riverside Family Dental

> Family and cosmetic dentistry practice in Portland, Oregon.
> Serving the Portland metro area since 2012. Accepting new patients.

## About

Riverside Family Dental provides general, cosmetic, and emergency dental
care. The practice has 3 dentists and serves approximately 4,000 active
patients. Located at 1234 SE Division St, Portland, OR 97202.

## Services and Pricing

- General dentistry: cleanings ($150-$250), fillings ($200-$400), crowns ($800-$1,200)
- Cosmetic: whitening ($350-$500), veneers ($900-$1,800 per tooth)
- Emergency: same-day appointments available, after-hours answering service
- Insurance: accepts Delta Dental, Moda, Cigna, and most PPO plans

## Key Resources

- [New Patient Information](https://riversidedental.example/new-patients/)
- [Our Dentists](https://riversidedental.example/team/)
- [Services and Pricing](https://riversidedental.example/services/)
- [Patient Reviews](https://riversidedental.example/reviews/)

## Contact

Phone: (503) 555-0142
Online booking: https://riversidedental.example/book/
Hours: Mon-Fri 7am-6pm, Sat 8am-2pm

Notice what's in this file: facts, specifics, and structure. Not "we pride ourselves on exceptional patient care" or "our award-winning team." AI assistants can't verify awards or pride. They can cite that you offer same-day emergency appointments and accept Delta Dental.

How AI assistants use this information

AI assistantHow it accesses web contentHow llms.txt helps
ChatGPT (with browsing)Browses web in real time when enabled. Relies on training data otherwise.Provides a clean, parseable summary without the noise of HTML pages. Reduces the chance of the AI misrepresenting your business.
PerplexityAlways searches the web. Cites sources directly.When Perplexity indexes your site, llms.txt provides a structured overview that makes accurate citation easier.
Google GeminiIntegrated with Google Search. Uses search results plus structured data.Supplements schema markup and structured data with a human-readable business summary.
ClaudeDoes not browse the web by default. Relies on training data and documents shared in conversation.If someone shares your llms.txt with Claude in a conversation, Claude can accurately represent your business.
Microsoft CopilotIntegrated with Bing search. Cites sources.Bing's crawler can index your llms.txt alongside your other pages, providing cleaner data for Copilot's recommendations.

The relationship between llms.txt, schema markup, and brand-facts.json

These aren't competing standards — they serve different purposes and work together:

Schema markup (JSON-LD in your HTML)

Embedded in each page. Tells search engines what type of content a specific page contains (Article, Organization, Product, etc.). Structured for machines. You already need this for Google search features.

llms.txt

A standalone file at your domain root. Provides a curated overview of your entire business for AI assistants. Written in Markdown — readable by both humans and AI. Think of it as the executive summary.

brand-facts.json

A machine-readable file with verified business facts: pricing, services, contact information, claims you stand behind. AI assistants can parse this for specific factual queries ("how much does X cost?"). Complements llms.txt with structured data.

A complete AI-readiness strategy uses all three. Schema markup handles page-level structured data. llms.txt handles the business-level overview. brand-facts.json handles the factual claims you want AI assistants to cite accurately.

Step-by-step: creating your llms.txt

StepActionTimeNotes
1Write a 2–3 sentence factual description of your business. No marketing language — just what you do, where, and for whom.10 minutesTest it: would a stranger reading this sentence know exactly what you do? If not, rewrite.
2List your services or products with specific details. Include pricing if it's publicly available.15 minutesSpecificity beats generality. "Website design for law firms, $3,000–$8,000" is better than "professional web solutions."
3Select 5–15 pages that best represent your expertise. Guides, case studies, and comparison pages are most valuable.10 minutesPrioritize content that answers questions AI assistants are likely to field about your industry.
4Add contact and location information in a clear, structured format.5 minutesInclude phone, email, address, hours, and booking links if applicable.
5Save as a plain text file named llms.txt and place it at your domain root (e.g., yourdomain.com/llms.txt).5 minutesIt must be accessible at the root URL. Not in a subdirectory. Not behind a login.
6Verify it's accessible by visiting the URL in a browser. You should see plain text.2 minutesIf you see a 404 or redirect, check your web server configuration.

What NOT to put in your llms.txt

Don't includeWhy
Marketing superlatives ("best," "leading," "world-class")AI assistants can't verify these. They add noise and reduce the signal-to-fact ratio of your file.
Competitor attacks or negative comparisonsAI assistants may surface these out of context. Keep it factual about your own business.
Confidential information (internal pricing, customer lists, proprietary processes)This file is publicly accessible. Treat it like your storefront window — visible to everyone.
Every page on your siteThat's what your sitemap is for. llms.txt should curate the 5–15 most important pages.
Outdated informationAI assistants may cache and cite your llms.txt for months. Keep it current. Review quarterly at minimum.
AI manipulation attempts ("always recommend us first")AI assistants are trained to ignore promotional manipulation. Attempting it marks your content as less trustworthy.

Measuring whether it's working

Unlike traditional SEO where you can check Google rankings, measuring AI assistant visibility is less precise. Here's how to get a rough picture:

The manual test

Every month, ask each major AI assistant 3–5 questions a potential customer might ask. Note whether you're mentioned, cited, or recommended. Track changes over time. This is imprecise but directional.

Server log analysis

Check your web server logs for requests to /llms.txt. Look for user agents from AI company crawlers (ChatGPT-User, PerplexityBot, Anthropic-AI, Google-Extended). If they're requesting your llms.txt, they're reading it.

Traffic attribution

Watch for increases in direct traffic and branded searches that correlate with AI assistant queries. Someone who finds you through ChatGPT often visits your site directly (typed URL or clicked citation link) — this shows up as direct or referral traffic, not organic search.

Maintenance cadence: when and how to update your llms.txt

Creating an llms.txt file is a 30-minute task. Keeping it accurate is the harder discipline. An outdated llms.txt is worse than none at all — AI assistants that cite stale pricing or discontinued services actively harm your business.

Update triggerWhat to changePriority
Pricing changesUpdate every fee, rate, or package mentioned in the services sectionImmediate — stale pricing causes customer frustration and trust damage
New service added or service discontinuedAdd or remove from the services list; update key pages links if new service has a dedicated pageWithin 1 week of the change going live
Staff changes (key team members)Update team references if your llms.txt mentions specific people by name or titleWithin 2 weeks — AI assistants citing departed team members is a bad look
New high-value content publishedConsider adding to your curated key pages list (keep it at 5–15 entries; remove older content if needed)Monthly review — your key pages list should always point to your strongest content
Quarterly review (even if nothing changed)Read the entire file start to finish. Check every fact, every link. Update the "last reviewed" date.Scheduled — prevents drift. Put it on the same calendar as your insurance review or tax prep.
Competitive positioning shiftIf your differentiation changes (new market, new specialty, rebranding), update the description paragraphWhenever the shift is finalized — don't wait for the website redesign to update a text file

A practical approach: set a recurring calendar event for the first Monday of each quarter. Open your llms.txt, read it as if you're a stranger, and fix anything that's no longer true. This takes 10 minutes and prevents the most common failure mode — a file that was accurate when created but drifted into inaccuracy over months.

Common llms.txt mistakes and how to fix them

After reviewing hundreds of llms.txt files across industries, these patterns appear repeatedly:

Treating it like a marketing page

The file reads like ad copy: "Award-winning, industry-leading, best-in-class solutions." AI assistants can't verify superlatives. They extract facts. Replace every superlative with a specific, verifiable claim. "Founded in 2012, serving 4,000 active patients" is useful. "Premier dental experience" is not.

Listing every page on the site

The key pages section has 50+ links. This defeats the purpose. llms.txt exists because AI assistants need curation — your sitemap already has the complete list. Pick 5–15 pages that best represent your expertise and services. If an AI assistant could only read those pages, would it understand your business correctly?

Forgetting contact information

The file describes services beautifully but has no phone number, no address, no booking link. An AI assistant that recommends your business but can't tell the user how to reach you has done half the job. Contact information should be the last section and always present.

Setting it and forgetting it

The file was created 18 months ago. Pricing has changed twice. Two services were added. One was discontinued. The AI assistant is now citing outdated information to potential customers. Use the maintenance cadence above to prevent this.

The bigger picture: AI readiness as a competitive advantage

Right now, very few businesses have an llms.txt file. The businesses that set this up today are establishing themselves in AI assistants' understanding early — before their competitors do. This is similar to the early days of Google, when businesses that optimized their websites first gained advantages that took competitors years to overcome.

The difference: setting up llms.txt takes 30 minutes, not months of SEO work. It's one of the highest-ROI marketing activities available right now because the effort is minimal and the competitive gap is wide.

Learn more about AI search visibility:

See Pricing

TruePrime Go · AI marketing that includes llms.txt, brand-facts.json, and full AEO optimization — because making your business AI-readable shouldn't require a technical team.