Software Services
For Companies
Products
Build AI Agents
Security
Portfolio
Build With Us
Build With Us
Top Software Developer
2026 Clutch Ranking
Top App Developer
2026 BusinessofApps
GEO and AI Search Visibility,
Built Into
Your Website
Generative engine optimization (GEO) is making your pages easy for ChatGPT, Gemini, Google AI Overviews and other AI engines to find, read and cite. Most of it is engineering: crawler access, llms.txt, Markdown versions of pages, structured data, answer-first pages and measurement. We built all of it on slashdev.io, and this page shows how.
What is GEO, and what does it take to get cited by AI engines?
GEO means making your pages easy for AI engines to retrieve, understand and quote. Start with what Google says: there are no special requirements for AI Overviews or AI Mode; a page must be indexed and eligible for a snippet, and the usual SEO basics apply. On top of that, six pieces of engineering help: let search and AI crawlers in and state your content-use preferences in robots.txt; publish an llms.txt file that lists your key pages; serve a clean Markdown version of each important page; add structured data that matches the visible text; write pages that answer the question first, with a review date and sources; and measure which AI answers cite you. None of it guarantees a citation. It removes the reasons an engine would skip you.
- What GEO is
- Making pages easy for AI engines to find, read and cite
- Google's position
- No special files or schema required for AI Overviews; indexed and snippet-eligible pages, SEO basics
- The engineering
- robots.txt signals, llms.txt, Markdown pages, structured data, answer-first pages, citation tracking
- Worked example
- slashdev.io, which runs every piece on this page
What GEO Engineering Covers
Each piece is a file, a header or a page pattern you can check, not a promise.
Crawler Access and Content Signals
Decide which crawlers may read the site and say how your content may be used. Cloudflare's Content Signals (search, ai-input, ai-train) and the IETF aipref draft's Content-Usage rule (search, ai-use, train-ai) both go in robots.txt. They state preferences; they do not block anyone.
- Search and AI crawlers allowed on purpose
- Content-Signal line in robots.txt
- Content-Usage line from the IETF draft
- Per-crawler rules where you need them
An llms.txt File
A Markdown file at /llms.txt that names the site, summarizes it and links the pages an agent should read first. The llms.txt proposal requires only the H1 title; the summary and link sections are optional.
- Site name and one-paragraph summary
- Sections of links with short notes
- Links to Markdown versions of pages
- An Optional section for secondary links
Markdown Versions of Pages
A clean text version of each important page, without navigation, scripts or styling, so an agent reads the content and nothing else. The llms.txt proposal suggests the same URL with .md added, and a rel="alternate" type="text/markdown" link to point to it.
- Same URL with .md added
- Accept: text/markdown served too
- Link header advertises the Markdown
- Built from the page's own copy only
Structured Data That Matches the Page
JSON-LD that describes the page, the service and the FAQs, using only what a reader can see. Google's guidelines say not to mark up content that is not visible, and that valid markup does not guarantee a rich result.
- JSON-LD, Google's recommended format
- One FAQPage and one BreadcrumbList per page
- WebPage and Service with a modified date
- Nothing in the markup that is not on the page
Answer-First Pages
Pages that state the answer in the first block, back it with a short list of facts, show when it was last reviewed and link the sources behind every third-party claim. Engines can quote a clear paragraph; they cannot quote a slogan.
- The question as a heading, the answer below it
- Facts in a short definition list
- A visible Last reviewed date
- Sources linked from the vendor's own docs
Measuring Citations in AI Answers
Track a fixed set of buyer questions in each AI engine on a schedule and record which sites each answer cites. Add the reports the engines publish: Google Search Console includes AI Overviews and AI Mode traffic in its Performance report, and Bing Webmaster Tools reports citations in Copilot and Bing AI summaries.
- A fixed list of buyer questions
- Which pages each engine cites
- Search Console and Bing AI Performance
- Changes tied to the pages that moved
From Audit to Measured Citations
Four steps, each with something you can check.
Audit Access and Pages
We read your robots.txt, sitemaps, rendering and structured data, and list the buyer questions you want to be cited for and which pages answer them today.
Fix the Plumbing
Content signals in robots.txt, an llms.txt, Markdown versions of key pages, and structured data that matches the visible text.
Rebuild Key Pages Answer-First
Pages that answer the question in the first block, with facts, a review date and sources, written in the words buyers use.
Measure and Adjust
We track which AI answers cite you for your list of questions, read Search Console and Bing AI Performance, and change the pages that are not cited.
What We Shipped on slashdev.io
The same pieces, running on this site. The code is copied from our own files.
Content Signals in robots.txt
Our robots.txt allows all crawlers and publishes our preferences in Cloudflare's Content-Signal format. We keep the IETF aipref Content-Usage line as a comment, because robots.txt validators such as Lighthouse reject it while the draft is unfinished. We chose to allow search, AI input and AI training, because we want AI engines to read and cite us. Your choice may differ, and both formats let you say no to training while allowing search.
- One group for all crawlers, no per-bot blocks
- Content-Signal: search, ai-input, ai-train
- Content-Usage (IETF draft) kept as a comment until validators accept it
- Sitemap listed for discovery
# public/robots.txt on slashdev.io (abridged)
User-agent: *
Allow: /
# AI content-use preferences
# (Cloudflare Content Signals + IETF aipref Content-Usage)
Content-Signal: search=yes, ai-input=yes, ai-train=yes
# Content-Usage: search=y, ai-use=y, train-ai=y (IETF aipref draft; kept as a
# comment because robots.txt validators such as Lighthouse reject the directive)An llms.txt That Points to Markdown
Our /llms.txt names the company, summarizes what we build and lists our key pages by section. Next to each page with a Markdown version, it lists that version too, and a longer /llms-full.txt sits in the Optional section.
- H1 title and blockquote summary
- Sections for services, AI agents and healthcare
- Pages paired with their Markdown versions
- Longer llms-full.txt marked optional
# public/llms.txt on slashdev.io (abridged)
# Slashdev (slashdev.io)
> A one-paragraph summary of what we build
## Healthcare software
- [Healthcare App Development](https://slashdev.io/healthcare-app-development): ...
- [Healthcare App Development (Markdown)](https://slashdev.io/healthcare-app-development.md): Markdown version of the page above, for AI agents.Our Guides Have Markdown Twins
Our service and healthcare guides on slashdev.io can be fetched as Markdown at the same URL with .md added, or by asking for text/markdown. The HTML page sends a Link header that points to the Markdown. The Markdown is built only from copy the HTML page already shows, and leaves out client names.
- /<page>.md for agents
- Accept: text/markdown on the normal URL
- Link rel="alternate" type="text/markdown"
- No text that the HTML page does not show
# Every page with a Markdown version answers in two ways
GET /healthcare-app-development.md
GET /healthcare-app-development
Accept: text/markdown
# and the HTML page advertises it:
Link: <https://slashdev.io/healthcare-app-development.md>; rel="alternate"; type="text/markdown"
Vary: Accept, Accept-EncodingGEO and AI Search Questions
What is generative engine optimization (GEO)?
GEO is the work of making your pages easy for AI engines such as ChatGPT, Gemini and Google AI Overviews to retrieve, understand and cite in their answers. It builds on SEO: an engine has to find and read a page before it can quote it.
Is GEO different from SEO?
It overlaps heavily. Google says there are no extra requirements for AI Overviews or AI Mode beyond being indexed and eligible for a snippet. The GEO-specific parts are for agents and other engines: llms.txt, Markdown versions of pages, content-use signals and measuring citations rather than only rankings.
Does Google need an llms.txt file?
No. Google says you don't need new machine-readable files or special schema to appear in AI Overviews or AI Mode. llms.txt is a proposal aimed at AI agents and tools that read a site at answer time; it costs little to publish and keep current.
Does a Content-Signal line in robots.txt stop AI crawlers?
No. Cloudflare describes content signals as preferences, not technical countermeasures, and says some companies may ignore them. The IETF aipref draft says the same of its Content-Usage preferences. To block a crawler, use crawler rules or bot management.
What is the difference between Content-Signal and Content-Usage?
Content-Signal comes from Cloudflare's Content Signals Policy and uses search, ai-input and ai-train. Content-Usage comes from the IETF AI Preferences (aipref) working group drafts and uses search, ai-use and train-ai. The aipref drafts are work in progress, not standards. We publish both lines.
Can I let ChatGPT search use my site but not train on it?
Yes. OpenAI documents separate crawlers: OAI-SearchBot for ChatGPT search and GPTBot for training, each controlled on its own in robots.txt. For Google, the Google-Extended token covers Gemini training and grounding, and Google says it does not affect inclusion or ranking in Google Search.
How do you measure citations in AI answers?
We run a fixed list of buyer questions in each AI engine on a schedule and record which sites each answer cites. We add Google Search Console, which includes AI Overviews and AI Mode traffic in its overall Performance report, and Bing Webmaster Tools' AI Performance report, which shows citations in Copilot and Bing AI summaries.
Can you guarantee our site will be cited?
No one can. Each engine decides what to cite. We remove the technical reasons to skip your pages, rebuild the pages that should answer your buyers' questions, and show you which answers cite you over time.
Documentation We Checked
The proposals, drafts and vendor pages behind the statements on this page.
- The /llms.txt filellms.txt proposal
- Giving users choice with Cloudflare's new Content Signals PolicyCloudflare blog, September 2025
- Managed robots.txt and Content SignalsCloudflare Docs
- A Vocabulary for Expressing AI Usage Preferences (draft-ietf-aipref-vocab)IETF Internet-Draft, work in progress
- Associating AI Usage Preferences with Content in HTTP (draft-ietf-aipref-attach)IETF Internet-Draft, work in progress
- AI features and your websiteGoogle Search Central
- General structured data guidelinesGoogle Search Central
- Google's common crawlers (Google-Extended)Google Search Central
- Overview of OpenAI crawlersOpenAI Docs
- Introducing AI Performance in Bing Webmaster ToolsBing Webmaster Blog, February 2026
Checked November 19, 2026.
Get Your Site Ready to Be Cited
Send us your site and the questions your buyers ask AI engines. We'll come back with what to fix, which pages to rebuild and a price.
About
Global Hubs
Engineering Skills
Social Media


@2026 slashdev.io