AEO Saigon
Fundamentals

AI Crawlers, robots.txt for AI, and llms.txt: Controlling Which Content AI Learns

Table of contents

In 2026, not just Googlebot but dozens of AI crawlers are continuously crawling the web to train AI models and provide real-time answers. Vietnamese businesses need to understand clearly: who is reading their content, where the control lies, and how to leverage this for AEO.

Major AI crawlers (2026)

CrawlerDeveloperUser-AgentPurpose
GPTBotOpenAIGPTBot/1.1Training + Browse Web
ClaudeBotAnthropicClaudeBot/0.1Training + web search
PerplexityBotPerplexity AIPerplexityBot/1.0Real-time search
Meta-ExternalAgentMetameta-externalagent/1.1Training Llama
BytespiderByteDanceBytespiderTraining + TikTok AI
Applebot-ExtendedAppleApplebot-Extended/0.1Apple Intelligence
Google-ExtendedGoogleGoogle-ExtendedTraining Gemini
CCBotCommon CrawlCCBot/2.0Open dataset (open-source training data)
AmazonbotAmazonAmazonbot/0.1Training Alexa/Bedrock
cohere-aiCoherecohere-ai/1.0Training API models

Controlling access with robots.txt

Blocking a crawler entirely

User-agent: GPTBot
Disallow: /

User-agent: ClaudeBot
Disallow: /

User-agent: Google-Extended
Disallow: /

Blocking by specific path (recommended)

# Cho phép AI crawl blog và trang marketing
User-agent: GPTBot
Allow: /vi/blog/
Allow: /vi/dich-vu/
Disallow: /api/
Disallow: /admin/
Disallow: /khach-hang/
Disallow: /bao-cao/
Disallow: /

User-agent: ClaudeBot
Allow: /vi/blog/
Allow: /vi/dich-vu/
Disallow: /

# Cho phép Perplexity (real-time search — cần để xuất hiện trong Perplexity)
User-agent: PerplexityBot
Allow: /
Disallow: /api/
Disallow: /admin/

# Chặn Meta và ByteDance (không muốn training Llama/TikTok AI)
User-agent: meta-externalagent
Disallow: /

User-agent: Bytespider
Disallow: /

Allowing all AI crawlers (AEO strategy)

# Cho phép toàn bộ AI crawlers — tối ưu cho AEO
User-agent: GPTBot
Allow: /

User-agent: ClaudeBot
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: Google-Extended
Allow: /

# Vẫn bảo vệ các path nhạy cảm
User-agent: *
Disallow: /api/
Disallow: /admin/
Disallow: /.env

The llms.txt standard

llms.txt is a file placed at the root of the domain (/llms.txt), helping AI understand the website's structure and content in a structured way — like sitemap.xml but for semantics.

Basic llms.txt structure

# AEO Saigon

> Agency chuyên AEO (Answer Engine Optimization) và JSON-LD schema cho doanh nghiệp Việt Nam muốn xuất hiện trong AI search và Google rich results.

AEO Saigon giúp SME tại TP.HCM và Hà Nội xây dựng hiện diện trong Google AI Overview, ChatGPT, Perplexity và Gemini thông qua structured data và content strategy.

## Tài liệu quan trọng

- [Hướng dẫn AEO cho SME Việt Nam](https://example.vn/vi/blog/aeo-guide): Tổng quan về Answer Engine Optimization và cách bắt đầu
- [JSON-LD Schema cơ bản](https://example.vn/vi/blog/json-ld-schema): Organization, LocalBusiness, FAQPage
- [Dịch vụ AEO](https://example.vn/vi/dich-vu/aeo): Gói dịch vụ và báo giá
- [Case Study](https://example.vn/vi/case-study): Kết quả thực tế từ khách hàng

## Điều kiện sử dụng

Nội dung blog có thể được AI trích dẫn với attribution. Không được sử dụng để training mô hình thương mại mà không có thỏa thuận bằng văn bản.

Complete llms.txt for businesses

# Tên Công Ty / Thương Hiệu

> [Mô tả ngắn 1-2 câu về công ty và giá trị cốt lõi]

[Đoạn mô tả chi tiết hơn, 3-5 câu, về sản phẩm/dịch vụ, thị trường phục vụ, và điểm khác biệt]

## Sản phẩm/Dịch vụ

- [Tên sản phẩm 1](URL): Mô tả ngắn
- [Tên sản phẩm 2](URL): Mô tả ngắn

## Tài liệu kỹ thuật

- [Tài liệu API](URL): Nếu có API public
- [Hướng dẫn tích hợp](URL)

## Blog & Tài nguyên

- [Tên bài blog chính 1](URL): Chủ đề
- [Tên bài blog chính 2](URL): Chủ đề

## Thông tin liên hệ

- Website: https://example.vn
- Email: contact@example.vn
- Địa chỉ: [Địa chỉ đầy đủ]

## Điều kiện sử dụng nội dung

[Nêu rõ: AI được phép/không được phép dùng nội dung này như thế nào]

Deploying llms.txt in Next.js

// app/llms.txt/route.ts
import { NextResponse } from "next/server";

export async function GET() {
  const content = `# AEO Saigon

> Agency chuyên AEO và JSON-LD schema cho SME Việt Nam.

AEO Saigon giúp doanh nghiệp xuất hiện trong Google AI Overview, ChatGPT, và Perplexity thông qua structured data strategy.

## Blog

${await getBlogList()}

## Dịch vụ

- [AEO Audit](https://aeosaigon.com/vi/dich-vu/aeo-audit): Kiểm tra và tối ưu hóa schema
- [Schema Implementation](https://aeosaigon.com/vi/dich-vu/schema): Triển khai JSON-LD đầy đủ
`;

  return new NextResponse(content, {
    headers: { "Content-Type": "text/plain; charset=utf-8" },
  });
}

async function getBlogList(): Promise<string> {
  // Lấy danh sách bài blog từ CMS/file system
  const posts = await getBlogPosts();
  return posts
    .map((p) => `- [${p.title}](https://aeosaigon.com/vi/blog/${p.slug}): ${p.description}`)
    .join("\n");
}

Decision matrix: Allow or block?

Content typeGPTBotClaudeBotPerplexityBotGoogle-Extended
Marketing blog, guides✅ Allow✅ Allow✅ Allow✅ Allow
Service pages, about pages✅ Allow✅ Allow✅ Allow✅ Allow
Public pricing✅ Allow✅ Allow✅ Allow⚠️ Depends
Product database (e-commerce)⚠️ Depends⚠️ Depends✅ Allow⚠️ Depends
Premium/paid content❌ Block❌ Block❌ Block❌ Block
Customer information❌ Block❌ Block❌ Block❌ Block
API endpoints❌ Block❌ Block❌ Block❌ Block
Admin/CMS❌ Block❌ Block❌ Block❌ Block

AI crawler checklist

  • Check server logs for AI crawler User-Agents
  • Review current robots.txt — are AI crawlers currently being blocked?
  • Decide on a strategy: allow, block by path, or block entirely
  • Create /llms.txt with an organization description and list of important pages
  • Ensure /api/, /admin/, and data paths all have Disallow for all bots
  • Verify robots.txt with Google Search Console > robots.txt Tester
  • Monitor traffic from AI crawlers monthly

Controlling AI crawlers is not about blocking everything or allowing everything. The right strategy is: allow AI access to content that carries brand value (blog, landing pages, service pages), block sensitive content and internal tools — then guide AI through llms.txt to ensure AI accurately understands your organization.

Frequently asked questions

How are AI crawlers different from Googlebot?

Googlebot crawls to index web pages for search results — you still control this via robots.txt and noindex. AI crawlers (GPTBot, ClaudeBot, PerplexityBot, Meta-ExternalAgent) crawl to collect training data for AI models or to provide real-time answers. Key difference: Blocking Googlebot affects traditional SEO. Blocking AI crawlers does NOT affect Google SEO — but it affects AI's ability to cite your content. Some AI systems (Perplexity, Bing Copilot) crawl in real-time when a question comes in — blocking these bots means AI won't find your latest content.

If I block GPTBot, will my content be removed from ChatGPT?

No — blocking GPTBot prevents OpenAI from crawling your site in the future, but does not delete data already collected. The training data for GPT-4o and earlier models has already been collected — blocking the bot now only affects future training runs and real-time web search (when ChatGPT Browse Web uses GPTBot). To request deletion of existing data: contact the AI developer directly via a privacy request — robots.txt has no effect on data already collected.

What is llms.txt and is it required?

llms.txt is a new standard (2025) placed at /llms.txt, allowing websites to instruct AI on how to use their content — similar to robots.txt but designed specifically for Large Language Models. The file contains: a brief description of the organization, a list of important pages and documents AI should read, and terms of content use. Not required, but major AI assistants (Claude, Perplexity) are integrating it to better understand websites. For Vietnamese businesses wanting AI to cite them accurately: llms.txt is a small investment with high ROI.

Should I allow or block AI crawlers?

It depends on your business goals. You should allow if: (1) you want AI to cite your content and brand when answering users, (2) you are building industry authority through AEO, (3) your content doesn't contain sensitive or proprietary information. You should consider blocking if: (1) you have a database or paid content you don't want AI to learn from, (2) commercially sensitive pages (pricing, contracts), (3) customer personal data (must block under PDPA/GDPR). Best strategy: block specific paths rather than blocking everything.

How do I check which AI crawlers have visited my site?

Check server logs (access logs) — filter by User-Agent containing 'GPTBot', 'ClaudeBot', 'PerplexityBot', 'Bytespider'. With Nginx: grep 'GPTBot' /var/log/nginx/access.log. With Google Search Console: crawl data only shows Googlebot, not other AI crawlers. With Cloudflare: Analytics > Bot Analytics shows traffic from known bots by category. Monitoring regularly helps identify new unknown crawlers and adjust robots.txt in time.

AEO Saigon

An Answer Engine Optimization agency in Ho Chi Minh City — helping business websites get cited by AI. About AEO Saigon →

Want your website to be cited by AI like this?

Free Audit