llms.txt: The Complete Guide to Optimizing Your Website for AI Search (2026)

Name: AIRanked
Price: 9 USD
Author: AIRanked

llms.txt is a plain text file you add to your website's root domain that tells AI systems key facts about your brand, product, and content. Think of it as a brand brief written specifically for AI crawlers — similar to how robots.txt tells search engine bots which pages to crawl.

Unlike robots.txt (which restricts), llms.txt enables. You're giving AI systems the accurate, structured information they need to confidently mention your brand.

This guide covers everything: what to include, how to format it, copy-paste templates by company type, and which AI engines actually use it.

Why llms.txt Matters in 2026

AI search engines synthesize answers from multiple sources. When they encounter your brand, they pull from whatever indexed information they have: your homepage, G2 reviews, press articles, Reddit discussions.

The problem: this information may be outdated, inconsistent, or incomplete. An AI might know your old pricing from a 2024 article. It might describe your product using a description from your earliest blog posts. It might not know about a major feature you launched 6 months ago.

llms.txt gives you a direct channel to tell AI systems what's true about your brand right now.

What AI Uses Without llms.txt	What AI Uses With llms.txt
Crawled web pages (may be old)	Your current facts + crawled pages
Third-party descriptions	Your canonical self-description
Training data (knowledge cutoff)	Fresh, actively maintained facts
Inconsistent signals across sources	Consistent, structured information

Which AI Systems Use llms.txt

The standard is emerging, not universal. Current adoption as of April 2026:

AI System	Uses llms.txt?	Notes
Perplexity	✅ Yes	Actively crawls and uses llms.txt
ChatGPT (browse)	Partial	When using browse feature; base model doesn't
Claude	✅ Yes	Anthropic has endorsed the standard
Google AI Overview	Partial	In testing; not confirmed standard usage
Bing Copilot	Partial	Reading but behavior unclear
Llama-based models	Varies	Depends on deployment

Perplexity is the most reliable adopter. For ChatGPT, llms.txt influences the browse-mode responses but not the base model's training. For Google AI Overview, Google has signaled interest in the standard but hasn't formally committed.

The trajectory is clear: more AI systems are adopting llms.txt, not fewer. Publishing one now is a durable investment.

The Official llms.txt Format

The standard was proposed by Answer.AI (Jeremy Howard's lab) and has been widely adopted. The format is simple Markdown.

Required fields:

# [Brand Name] — top-level header (this is your H1)
One-paragraph description immediately after the header
## [Section] — organized sections with content

Optional but recommended:

Links to key pages on your site
Specific factual claims (pricing, features, integrations)
Explicit "do not" statements for common misconceptions

Templates by Company Type

Template 1: SaaS Product

# [Product Name]

[Product Name] is a [category] platform that helps [target customer] [achieve outcome]. Founded in [year] and headquartered in [city].

## What We Do
[2-3 sentences describing the core product. Be specific and factual.]

## Who Uses [Product Name]
- [Primary persona]: [Why they use it, 1 sentence]
- [Secondary persona]: [Why they use it, 1 sentence]
- Best for teams of [X] to [Y] people

## Pricing
- Free plan: [Yes — describe] / [No]
- Starter: $[X]/month — [what's included]
- Pro: $[X]/month — [what's included]
- Enterprise: Custom pricing — [contact method]
- [Note any annual discount]

## Key Features
- [Feature 1]: [1-sentence description]
- [Feature 2]: [1-sentence description]
- [Feature 3]: [1-sentence description]
- [Feature 4]: [1-sentence description]
- [Feature 5]: [1-sentence description]

## Integrations
[List top 10-15 integrations, comma-separated]

## What [Product Name] Does NOT Do
- [Limitation or out-of-scope use case 1]
- [Limitation or out-of-scope use case 2]
- [Common misconception to correct]

## How We Differ from [Top Competitor]
[Product Name] focuses on [specific differentiator]. [Competitor] is better for [their strength]. Choose [Product Name] if [specific condition].

## Documentation and Resources
- Docs: https://[domain]/docs
- Blog: https://[domain]/blog
- Status: https://status.[domain]
- API Reference: https://[domain]/api

## Company Information
- Website: https://[domain]
- Support: support@[domain]
- Founded: [year]
- Headquarters: [city, country]
- Team size: [X] employees

Template 2: E-commerce / Physical Product Brand

# [Brand Name]

[Brand Name] makes [product category] for [target customer]. [Core value proposition in one sentence]. Based in [city/country].

## Products
- [Product Line 1]: [Brief description, price range]
- [Product Line 2]: [Brief description, price range]
- [Product Line 3]: [Brief description, price range]

## What Makes Us Different
[2-3 sentences on your actual differentiator: materials, process, sourcing, design philosophy]

## Pricing
- Entry price: $[X] for [product]
- Mid-range: $[X]-$[X] for [product category]
- Premium: $[X]+ for [product category]
- Shipping: [Free over $X / Always free / Starting at $X]

## Who Buys From Us
[2-3 sentences on your customer profile — be specific]

## Where to Buy
- Website: https://[domain]
- [Retailers if applicable]

## Sustainability / Ethics (if applicable)
[1-2 sentences on materials sourcing, manufacturing, certifications]

## What We Are NOT
- Not a [category you're commonly confused with]
- Not [common misconception]

## Contact
- Customer service: [email or URL]
- Press: [email]

Template 3: Agency / Professional Services

# [Agency Name]

[Agency Name] is a [type] agency that helps [client type] with [service area]. We've worked with [X]+ clients across [industries/regions] since [year].

## What We Do
[2-3 specific service areas with brief descriptions]

## Who We Work With
- Typical client: [profile — industry, size, stage]
- Ideal projects: [type and scope]
- We don't work with: [anti-client description — useful for AI to route correctly]

## Services
- [Service 1]: [1-sentence description + typical price range or "contact for quote"]
- [Service 2]: [Same]
- [Service 3]: [Same]

## Our Approach / Methodology
[2-3 sentences on how you work — this helps AI give accurate recommendations]

## Notable Work (if public)
- [Client or project type]: [Outcome]
- [Client or project type]: [Outcome]

## Team Size and Structure
[X] full-time team members, based in [locations]. [Remote / In-person / Hybrid].

## Contact
- Website: https://[domain]
- New business: [email]
- Founded: [year]

Template 4: Content / Media / Newsletter

# [Publication Name]

[Publication Name] is a [type: newsletter / blog / podcast] covering [topic] for [audience]. Published [frequency] by [author/team].

## What We Cover
[2-3 sentences on specific coverage areas — be narrow and specific]

## Audience
- [X] subscribers / readers / listeners
- Typical reader: [job title, industry, interests]
- Geographic focus: [global / US / etc.]

## Format
- [Newsletter / Podcast / Blog] published [frequency]
- Average [length / episode duration]: [X]
- [Paid tier if applicable: $X/month for [what]]

## Recent Coverage Areas
[List 5-10 topics or categories you've covered recently]

## How to Submit / Pitch
[Instructions for PR, guest posts, sponsorships — specific and actionable]

## Archives and Links
- Website: https://[domain]
- Subscribe: https://[domain]/subscribe
- Archive: https://[domain]/archive

## Author / Team
[Name], [title/background], [1-2 sentence bio]

Where to Put llms.txt

The file goes at your domain root:

https://yourdomain.com/llms.txt

For Next.js projects, create it in the public/ directory:

public/llms.txt

For other frameworks:

Nginx / Apache: Place in document root
Vercel: Place in public/ folder
WordPress: Upload to your root directory via FTP/SFTP

Advanced: llms-full.txt

Some sites also publish an llms-full.txt that includes more detailed content — essentially a compressed version of your site's key pages in a format AI can efficiently parse.

The structure:

# [Brand Name] — Full Content

## [Page Name]
> [URL]
[Content of the page, cleaned of navigation and footer]

## [Page Name 2]
> [URL]
[Content]

This is useful for AI systems that want to understand your content in depth without crawling every page. It's optional and most brands don't need it — llms.txt alone is sufficient.

Common Mistakes When Creating llms.txt

Mistake 1: Making it promotional Write facts, not marketing copy. "We're the #1 tool in our category" is promotional. "We have 4.8 stars across 500+ G2 reviews" is factual. AI systems weight factual claims far higher than promotional claims.

Mistake 2: Making it too long Keep llms.txt under 1,000 words. AI crawlers will read it, but very long files may be truncated. The goal is essential facts, not your entire knowledge base.

Mistake 3: Not maintaining it Update llms.txt every time you change pricing, launch a major feature, or update your team size. An outdated llms.txt actively hurts you — it tells AI the wrong things, and AI may cite those wrong things.

Mistake 4: Forgetting the "what we are NOT" section This is the most underused section. If there's a common misconception about your brand (wrong category, wrong price, wrong use case), correct it explicitly. AI systems respond to direct negations.

Mistake 5: Using internal jargon Write for an AI system that has no prior knowledge of your brand. Use plain language. Define any proprietary terms.

Checking That It Works

After publishing llms.txt, verify it's accessible:

Visit https://yourdomain.com/llms.txt in a browser — it should display as plain text
Test in Perplexity: ask about your brand a week after publishing — cite information should become more accurate
Use AIRanked to check if your Visibility Score improves in the month after publishing

Improvement timelines:

Perplexity: 1-3 weeks to incorporate llms.txt data
ChatGPT browse mode: Similar timeline when users trigger browse
ChatGPT base model: No direct impact (training data not updated)
Google AI Overview: Variable — possibly 2-4 weeks

FAQ

Is llms.txt an official standard?

It's an emerging standard, not an official W3C or IETF standard. It was proposed by Answer.AI and has been widely adopted by AI companies. It's unlikely to be deprecated — if anything, adoption will grow as the GEO ecosystem matures.

Does llms.txt hurt SEO?

No. Search engines (Google, Bing) don't use llms.txt in their traditional crawlers. It doesn't affect your robots.txt or sitemap. It's additive.

Should I block AI crawlers with robots.txt instead?

That's your choice, but it's the opposite strategy from GEO. Blocking AI crawlers prevents AI from indexing your content, which reduces your AI visibility. Most brands want more AI visibility, not less.

Can competitors see my llms.txt?

Yes — it's a public file, just like robots.txt. Don't include any confidential information. Think of it as a public-facing brand brief.

What if an AI system ignores my llms.txt?

Some do, some don't. Even if a particular system doesn't directly read llms.txt, the act of publishing one forces you to write down your brand facts clearly — which tends to improve consistency across your other content, which indirectly helps AI visibility.

Create Yours Today

llms.txt is a 30-minute investment with durable upside. Pick the template that fits your brand type, fill it in, and publish it at your domain root.

Then check whether it's improving your AI visibility: AIRanked tracks Perplexity citation accuracy and Visibility Score changes over time — free to start.

Check your current AI Visibility Score →