How much does llms.txt cost?

If the website already has a CMS, modifying or adding a Markdown file costs almost nothing. The technical difficulty is low - an average web developer can complete it in 1-2 hours.

I have a Facebook Page, is that enough?

AI platforms cannot directly read Facebook Posts. llms.txt works on your brand website, enabling search engines like ChatGPT to understand your brand content. They complement each other.

When will it take effect?

Unlike SEO which takes months to show results, after AI crawlers read your llms.txt, the next time a user asks 'Which hotel in Macau is good?', your brand could be included in the answer. The timing is determined by the algorithm and cannot be precisely predicted.

How should I write it?

Use Markdown format to list your brand positioning, core services, FAQ, pricing, and booking methods. First list 10-20 items you want AI to recommend, then ask a developer to create the file.

Do restaurants need it too?

Especially important for F&B businesses. When tourists ask 'Which Portuguese restaurant in Macau is good?', restaurants with llms.txt are more likely to be recommended by AI. It's recommended to include signature dishes, business hours, address, and phone number.

How is it different from SEO?

SEO targets traditional search engines like Google; llms.txt targets AI assistants like ChatGPT and Perplexity. The technologies are different and target user behaviors vary as well, so it's recommended to deploy both.

Will Perplexity use it?

Yes, it will. AI search engines like Perplexity need to read website content to generate answers. llms.txt directly tells them which pages are most valuable, essentially helping your brand 'get on the list'.

What if I don't have a technical team?

You can ask a web developer for assistance. Generally, you just need to copy the core content from your existing website and create a Markdown format file placed in the root directory - the process is not complicated.

Will I definitely be recommended after creating it?

Not guaranteed. The purpose of llms.txt is to improve AI readability, making it easier for AI to understand your brand, but whether you're ultimately recommended depends on AI's comprehensive judgment.

Do hotels and souvenir shops need it too?

Yes, both need it. Tourists often ask 'Which souvenir shop in Macau is good?' or 'Which hotel is near the Ruins of St. Paul's'. llms.txt helps AI answer these questions more accurately and increases your chances of being selected.

llms.txt: The AI Traffic Secret Weapon Ignored by 95% of Brands (with Triangle IP 5x Growth Verification) — Global Business Insights | CloudPipe AI

This article analyzes how llms.txt helps Macau food and tourism businesses improve AI search visibility, covering technical principles, Triangle IP case studies, JSON-LD collaboration, 48-hour deployment process, and CloudPipe continuous management services, helping businesses get more accurate recommendations on ChatGPT, Perplexity, Gemini, and Copilot.

What Is llms.txt? Why It Is the Most Underrated Technology of 2025

llms.txt can be understood as “a website guide written for AI.” If robots.txt tells crawlers “where not to go,” and sitemap.xml lists “all pages on a website,” then llms.txt tells ChatGPT, Perplexity, Gemini, and Copilot: “which pages are most worth reading, and how to understand this brand.”

Technically, it is a Markdown file placed in the website root directory, such as example.com/llms.txt, used to organize brand positioning, core services, FAQs, menus, room types, transport information, booking pages, price lists, and high-value content. According to Codersera’s 2026 guide, llms.txt was proposed by Jeremy Howard in 2024. A study covering 300,000 domains found that only around 10% of websites have deployed llms.txt, meaning most brands have not yet built AI-readable infrastructure.

The point is not to “make AI rank you immediately,” but to first present your most important content in a way that AI can easily understand.

For restaurants, hotels, souvenir shops, and tourism businesses in Macau, the first step is simple: list the 10 to 20 pages you most want AI to recommend, such as Portuguese restaurants, family-friendly restaurants, souvenir shops, guided tour services, hotel packages, transport guides, booking pages, and frequently asked questions. Add a one-sentence description next to each link to clearly explain what question that page answers.

ChatGPT tends to provide synthesized answers; OpenAI DevDay 2025 announced that ChatGPT has over 800 million weekly users.
Perplexity tends to cite sources; its CEO said it handled 780 million queries in May 2025.
Gemini is connected to the Google Search ecosystem; Google AI Overviews has reached 2 billion monthly users, while the Gemini App has around 450 million monthly active users.
Copilot is often connected to Bing and business search scenarios, and Microsoft also emphasizes that Copilot Search provides source citations.

In practice, llms.txt should be implemented together with content quality, Schema, reviews, external mentions, and technical SEO. It is not magic, but as AI search becomes a mainstream entry point, it is one of the lowest-cost and earliest foundational upgrades Macau SMEs can put in place.

Sources: Codersera llms.txt Guide, OpenAI DevDay 2025, TechCrunch / Perplexity Query Volume, TechCrunch / Google AI Overviews, Microsoft Copilot Search

Triangle IP Case Deep Dive: How llms.txt + JSON-LD Drove 5x AI Traffic

Triangle IP is a patent management SaaS platform for inventors, patent attorneys, and enterprise IP teams. Its approach was not simply to “publish more articles,” but to organize its most commercially valuable pages into AI-readable entry points: using llms.txt to mark high-priority content such as Blog, Use Cases, Pricing, Product, and Docs, making it easier for tools like ChatGPT, Perplexity, Gemini, and Copilot to determine “which pages are worth citing.” A public case study from Concurate shows that after implementation, one Triangle IP article saw AI-driven data increase from 17 sessions and 16 events to 23 sessions and 99 events, representing roughly 5.5x growth in AI-triggered interactions. Source: Concurate Triangle IP llms.txt Case Study.

The key point is not whether AI will “automatically reward” llms.txt, but whether you proactively tell AI: these pages represent my brand, products, pricing, use cases, and expert answers.

If llms.txt is an “AI navigation guide,” then JSON-LD is an “AI identity card.” The former lists the pages worth reading; the latter uses Schema.org structured data to define entity relationships such as company, product, FAQ, service area, reviews, and pricing. For Macau businesses, this is especially important: restaurants, wholesalers, clinics, tutoring centers, and engineering companies should not simply wait for AI to infer what they do from webpage text. Instead, they should use JSON-LD such as Organization, LocalBusiness, Product, Service, and FAQPage to state it clearly. Structured data does not guarantee citation, but it can reduce AI misunderstanding of brand positioning, which also aligns with multiple GEO technical guidelines recommending “machine-readable content.”

This case should also be understood within the broader trend. Adobe Analytics reported that in February 2025, traffic to U.S. retail websites from generative AI sources had grown 1,200% compared with July 2024, and had roughly doubled every two months since September 2024. Statcounter’s 2025 data also shows that among AI chatbot referrals, ChatGPT accounted for 79.8%, Perplexity for 11.8%, Microsoft Copilot for 5.2%, and Google Gemini for 2%. Sources: Adobe Analytics, Statcounter.

How Can Macau Businesses Apply This?

Step 1:Create /llms.txt and include only the 10 to 30 most important URLs, such as service pages, pricing pages, FAQs, case studies, About Us, and contact pages.
Step 2:Write one clear description for each URL, specifying the target audience, location, use case, and differentiated selling points.
Step 3:Add JSON-LD to core pages, prioritizing Organization, LocalBusiness, Service, and FAQPage.
Step 4:Use GA4 every month to filter referrals from chatgpt.com, perplexity.ai, gemini.google.com, and copilot.microsoft.com, then track AI traffic, time on page, and inquiry conversions.

Anatomy of an Effective llms.txt: Structure, Priorities, and Best Practices

An effective llms.txt is not about dumping every URL on your site into a text file. It is about giving AI assistants such as ChatGPT, Perplexity, Gemini, and Copilot a brand entry point that is “citable, understandable, and verifiable.” According to SE Ranking’s analysis of around 300,000 domains, only 10.13% of websites have an llms.txt file, and there is currently no clear evidence that the file alone can increase AI citation rates. In other words, the opportunity lies in “building low-cost infrastructure,” but it should not be seen as a replacement for content and Schema. Source: Search Engine Journal / SE Ranking.

For SMEs in Macau, the value of llms.txt is not as a “magic ranking file,” but as a way to proactively prioritize the pages that AI should understand about your business.

Basic Structure: Start with a Brand Summary

The first paragraph should use 2 to 3 sentences to clearly explain who you are, whom you serve, what problem you solve, and which region you serve. For example, Triangle IP’s llms.txt sample begins with one sentence describing its patent management platform, then groups links such as Blog, Use Cases, Pricing, and Product, so AI does not need to infer priorities from the navigation menu. Source: Concurate Triangle IP example.

Content Priorities: Put Conversion-Critical Pages First

First priority:Product / service pages, pricing pages, booking pages, or enquiry pages, because these directly affect whether AI recommends you.
Second priority:Use cases, FAQs, comparison pages, and industry guides, which help Perplexity and Copilot answer high-intent questions such as “which provider is right for me?”
Third priority:Blogs, news, and case studies, which should be used to build professional context rather than occupy the top positions.

Implementation Advice: llms.txt Should Work Together with JSON-LD

The best practice is: llms.txt handles navigation, while JSON-LD provides proof. For example, a restaurant in Macau can include entry points such as “menu, reservations, family-friendly dining, and private events” in llms.txt, while also adding LocalBusiness, FAQPage, Product, or Service Schema to its pages. The AI Search market is no longer limited to ChatGPT; 2026 data shows that ChatGPT accounts for around 64.5%, Gemini around 21.5%, Perplexity around 6.4-8%, and Copilot around 2%. This means structured content should support the reading habits of multiple platforms. Source: theStacc AI Search Market Share 2026.

Actionable checklist: Review llms.txt once a month; remove outdated links; keep each section to a maximum of 5 to 8 high-value URLs; add a one-sentence purpose description after each link; and make sure robots.txt is not blocking GPTBot, PerplexityBot, Google-Extended, or Bingbot. Otherwise, no matter how well written your AI entry point is, it may not be readable by AI systems.

The Synergy Between llms.txt and JSON-LD Schema

If llms.txt is a “navigation map” for AI, then JSON-LD Schema is the “ID card” for each important page. The former tells ChatGPT, Perplexity, Gemini, and Copilot which pages are worth understanding first; the latter uses structured data to specify whether a page represents a company, service, FAQ, product, address, or review, making it easier for AI to judge whether the content is trustworthy and cite-worthy.

The key point is not that “adding llms.txt will instantly drive traffic.” After analyzing around 300,000 domains, SE Ranking found that only 10.13% of websites had an llms.txt file, and there was no clear evidence that the file alone increased AI citation rates. Google has also stated that JSON-LD is its recommended structured data format, but correct markup does not guarantee rich results. Sources: Search Engine Journal / SE Ranking, Google Search Central

Why Should Both Be Implemented Together?

Because the four major AI entry points source information differently: ChatGPT Search uses web sources and search partners; Copilot is closely tied to the Bing ecosystem, and Bing Webmaster Tools already supports JSON-LD validation; Gemini is connected to Google Search; while Perplexity places particular emphasis on citable sources. In other words, llms.txt helps guide AI to the “right pages,” while JSON-LD makes the page’s entities, services, FAQs, and location information clearer.

How Can Macau Businesses Put This Into Practice?

Include only high-value pages in llms.txt: the homepage, service pages, FAQs, case studies, and contact page. Do not dump every site URL into the file.
Add JSON-LD to every core page: use LocalBusiness for local shops, Organization + Service for B2B services, and FAQPage for frequently asked questions.
Content and Schema must be consistent: Google clearly requires structured data to represent visible page content. Do not mark up reviews, prices, or services that do not exist on the page.
Track performance with Search Console, Bing Webmaster Tools, and AI referrals: do not look only at rankings. Record whether ChatGPT, Perplexity, Gemini, and Copilot start citing your brand pages.

For SMEs, the best approach is to treat llms.txt as the AI navigation layer and JSON-LD as the trust and semantic layer. The lesson from cases such as Triangle IP’s 5x growth is not that a single file can do everything, but that clear content architecture, verifiable data, case study pages, and structured markup work together to improve AI readability.

48-Hour Deployment Guide: From Zero to an Effective llms.txt

Deploying llms.txt does not require a major website overhaul. The key is to create an “AI-readable content entry point” within 48 hours, then pair it with JSON-LD Schema so ChatGPT, Perplexity, Gemini, and Copilot can more easily understand your brand. Note: llms.txt is still an emerging convention, and not every AI platform guarantees that it will read it directly. However, SE Ranking’s analysis of around 300,000 domains shows that only 10.13% of websites have implemented it, which means Macau businesses still have an early-mover opportunity.

Day 1: Organize the Pages AI Should Read First

List 10 to 20 core URLs:company profile, service pages, product pages, FAQ, case studies, pricing, store addresses, and contact details.
Write a one-sentence summary for each URL:do not stuff keywords; clearly explain “what question this page answers.”
Check Schema at the same time:use Organization for company pages, LocalBusiness for local merchants, FAQPage for FAQ pages, and add Product or Service to product or service pages.

Day 2: Publish, Test, and Track

Place the file in the root directory:ensure it can be opened at /llms.txt, using a clean Markdown format.
Test it on the four major platforms:ask ChatGPT, Perplexity, Gemini, and Copilot, “What services does this company provide?” and check whether the answers are accurate.
Set up tracking:monitor referrals from chatgpt.com, perplexity.ai, copilot.microsoft.com, and gemini.google.com in GA4. In the Triangle IP case study, Concurate stated that its AI tool traffic delivered visible gains and generated 500+ signups.

Practical advice for Macau businesses:do not aim for perfection at the start. Complete the first version within 48 hours, then update it once a month based on new services, FAQs, media coverage, and customer case studies, so AI can continuously read the latest and most credible information about your brand.

CloudPipe llms.txt Management Service: Ongoing Optimization, Not One-Time Deployment

llms.txt is not an SEO file that you “set once and forget.” It is an ongoing operational entry point for AI visibility. SE Ranking’s analysis of around 300,000 domains shows that only 10.13% of websites have deployed llms.txt, and there is currently no clear correlation with AI citation rates (source: Search Engine Journal / SE Ranking). This means Macau businesses should not focus only on whether the file exists, but should regularly update brand information, service pages, FAQs, case studies, Schema, and internal links, giving ChatGPT, Perplexity, Gemini, and Copilot a clearer path to understanding the brand.

Practical takeaway: llms.txt is an AI content map, not a guaranteed ranking tool; its real value comes from continuously organizing the content you want AI platforms to cite.

CloudPipe’s Approach

Monthly review:Update the core pages, service descriptions, case study links, and outdated information within llms.txt.
Platform coverage:Track citation changes across ChatGPT, Perplexity, Gemini, and Copilot for brand names, service keywords, and location-based keywords.
Structured support:Maintain llms.txt together with JSON-LD Schema, FAQPage, Organization, and Service markup.
Performance validation:Using Triangle IP’s case study of around 5x growth from AI-readable content and search visibility optimization as a reference, include AI citations, crawler visits, and query rankings from 14 to 30 days after deployment in reporting.

Macau SMEs are advised to start with the 10 to 20 most important pages: homepage, service pages, pricing or plans, FAQs, customer case studies, and location landing pages. CloudPipe turns these pages into AI-friendly content entry points and adjusts them monthly, instead of delivering a static file and stopping there.

llms.txt: The AI Traffic Secret Weapon Ignored by 95% of Brands (with Triangle IP 5x Growth Verification)