More people now ask an AI assistant instead of typing a search: “what is the best template for a designer portfolio?” or “how do I build a landing page with Claude Code?” When the answer names a website, that website gets visitors and trust. So it is natural to ask how to become one of the sites the AI mentions.

The honest answer comes in two parts. Nobody can pay or trick an AI assistant into recommending a site, and anyone who promises it is selling something. But you can remove the obstacles that keep assistants from finding and using your pages, and you can create the kind of content they are most likely to cite. That is what this guide covers.

How AI assistants decide what to mention

An assistant can answer in two ways, and they work differently.

From what it learned in training. The model has absorbed a large amount of text from the web up to a cutoff date. Being mentioned this way depends on how often and how clearly your site and brand were discussed across the web before that date. It is slow, indirect and mostly outside your control.

From a live search. Many assistants can search the web when a question needs fresh or specific information, then read pages and cite them. ChatGPT, Claude, Perplexity, Microsoft Copilot and Google’s AI features all have some version of this. This is where you have the most influence, because you can make your pages easy to find, fetch and quote.

Different assistants use different sources. Some rely on their own crawler, some on a search partner’s index and some on a mix. The practical consequence is that you should make your site available everywhere rather than guess which index one assistant uses today.

Step 1: Let the crawlers in

If an assistant’s crawler cannot fetch your pages, nothing else matters. Check two places.

Your robots.txt. This file at yourdomain.com/robots.txt tells crawlers where they may go. Each company runs separate crawlers for separate jobs, and you can treat them differently. These are the main ones at the time of writing, so confirm in each company’s own documentation, since names change:

  • OpenAI: OAI-SearchBot finds pages for ChatGPT search, GPTBot is used for training models and ChatGPT-User fetches a page when a user asks about it.
  • Anthropic: Claude-SearchBot helps with search results, ClaudeBot is used for training and Claude-User fetches pages on a user’s request.
  • Perplexity: PerplexityBot indexes pages for its search results.
  • Google: Googlebot handles Google Search. Google-Extended is a control token for whether your content is used for Google’s Gemini models. It does not affect your place in Google Search.
  • Microsoft: Bingbot powers Bing, whose index feeds some other tools.

Blocking search crawlers, such as OAI-SearchBot or Claude-SearchBot, can reduce your visibility in those assistants’ answers. Blocking training crawlers is a separate choice. That decision is yours, and it is a trade-off between wanting to be recommended and not wanting your content used to train models.

The simplest setup, if you want maximum visibility, is a robots.txt that does not block any of them:

User-agent: *
Disallow: /api/
Disallow: /private/

Sitemap: https://yourdomain.com/sitemap.xml

Your firewall or CDN. Services such as Cloudflare can block bots automatically, and some have switches for blocking AI crawlers. If you did not set them on purpose, check that they are off. A simple test: request your site with an AI crawler’s name in the user agent and confirm you get a normal page, not an error or a challenge.

Step 2: Get indexed in the main search engines

Assistants that search the web depend on indexes. If your pages are not in Google or Bing, they are much harder to find. Do this once:

  1. Verify your site in Google Search Console and Bing Webmaster Tools.
  2. Submit your sitemap.xml in both.
  3. Request indexing for your most important pages.
  4. Make sure no noindex tag is left on the site.

Bing’s tool can import your site straight from Google Search Console, so the second one takes a few minutes. It also now includes an AI performance report in beta that shows when your pages are cited in AI answers, which is worth checking over time. Our SEO checklist for AI-built websites lists the technical basics.

Step 3: Write pages that answer questions

This is the part with the biggest effect. Assistants quote pages that state an answer plainly. A few habits make your pages easier to use:

  • Put the answer first. Under each question-style heading, give the answer in the first sentence or two, then add detail.
  • Use headings that match real questions. “How much does a portfolio website cost?” is more useful than “Pricing thoughts”.
  • Define terms. A sentence such as “An AI-ready website template is a template organized so that an AI agent can read, change and check it” is easy to lift into an answer.
  • Be specific. Names, numbers, steps and examples are what makes a source worth citing. General advice that fits every website is not.
  • Add a short FAQ to pages where people have clear questions, and mark it up with FAQ structured data.
  • Show dates and keep content current. Assistants favor information that looks maintained. Update a page when something changes and say when it was last updated.

Do not write for the machine at the expense of the reader. Pages stuffed with repeated keywords or hundreds of near-duplicate pages are a risk with search engines and are less useful to assistants too.

Step 4: Make the page easy to parse

Assistants and crawlers read the HTML, not the picture of the page.

  • Put your main content in the HTML, not only inside images or content that loads after a click.
  • Use one <h1>, ordered headings, lists and short paragraphs.
  • Add structured data (JSON-LD) for your organization, website, articles, products and FAQs, so the facts are stated explicitly.
  • Keep the page fast and the markup clean, since crawlers have limited time per site.

Some sites also publish an llms.txt file, a short Markdown summary of the site with links to its key pages. It is a proposed convention, and the major assistants have not confirmed that they rely on it, so treat it as an inexpensive extra, not a ranking factor.

Step 5: Be mentioned elsewhere

Assistants lean on sources they consider trustworthy, and they notice what other sites say about you. Appearing in the places your audience reads helps in two ways: it brings direct visitors, and it gives assistants more independent evidence that you exist and what you do.

Useful places depend on your field, but common ones include:

  • Product directories, marketplaces and “awesome” lists in your niche
  • GitHub, if you publish anything open source or code related
  • Communities such as Reddit, Discord servers and forums, where you take part honestly and do not just post links
  • Guest articles, podcasts and interviews
  • Reviews on independent sites

Use the same short description of what you do everywhere, so your name becomes associated with a clear topic. For example, we describe Weblisy as “AI-ready Astro templates and component prompts”.

Step 6: Measure and be patient

Results take weeks or months. You can track progress without special tools:

  • Ask the assistants. Every couple of weeks, ask ChatGPT, Claude and Perplexity the questions your customers would ask, and note whether your site appears and which page is cited. Keep the list of questions the same, so you can compare over time.
  • Check your analytics for referrals. Visits from chatgpt.com, perplexity.ai and similar sources show that assistants are sending people to you.
  • Watch the Search Console and Bing reports for indexing and impressions.

Answers vary between sessions and users, so look for a trend, not a single result.

What does not work

  • Paying for a “guaranteed AI ranking”. There is no such product.
  • Hidden text aimed at AI crawlers. Search engines treat hidden text as spam, and assistants can read what visitors see.
  • Mass-produced pages. Publishing hundreds of thin pages for every possible question lowers the quality of your site.
  • Fake reviews or fake mentions. They are easy to detect and damage trust.

A short checklist

  1. Crawlers are not blocked in robots.txt or by your firewall.
  2. The site is verified in Google Search Console and Bing Webmaster Tools, with the sitemap submitted.
  3. Each key page answers a clear question in its first lines.
  4. Structured data matches what is on the page.
  5. The site is mentioned on other relevant sites, described consistently.
  6. You check the assistants every few weeks and adjust.

Build the foundation with your own site

If you are starting a new site, a clean structure gets most of these steps done for you. Every Weblisy template is an Astro project, which produces fast, crawlable HTML by default, and our guide to AI-ready templates shows what makes a project easy for both people and AI to work with. To write pages that are specific enough to be cited, How to write AI prompts that build better websites is a good next read.