What We Do for AI Search Visibility, and What We Do Not

People increasingly ask AI assistants questions they used to type into a search engine. Organizations naturally want to be part of those answers, and a whole industry has appeared promising to make that happen, sometimes under names like generative engine optimization.

Some of that advice is sound. Some of it is wishful thinking. Some of it is manipulation that search engines explicitly prohibit. Here is how we approach AI search visibility for our own sites and for the websites on our platform, including what we deliberately do not do.

What genuinely helps

The foundations of being found by AI assistants are the same as being found by people:

• Pages that can be crawled and read. Content that loads as real text, with clear headings and a sensible structure, is easier for any system to understand.
• Useful, specific content. Pages that answer real questions with real detail are worth citing. Thin pages written only to rank are not.
• Accurate structured data. Machine-readable descriptions of articles, authors, organizations and visible FAQs help systems understand what a page is. We explain our rules in structured data that matches the page.
• A complete sitemap. A sitemap lists every page worth finding, including blog posts and their images.
• Clear identity. Consistent organization details, social profiles and author pages help systems connect content to the people behind it.

None of these are tricks. They are what a well-built website should have anyway.

Crawler controls with honest labels

Different AI companies send different crawlers for different purposes. Some fetch pages for search answers. Some fetch a page when a person asks about it. Some collect material for training models. Our search settings group these crawlers by purpose so organizations can decide which to allow, with plain explanations of what each switch does and does not control.

We are careful to describe these switches honestly. Allowing a search crawler does not guarantee that a page will be fetched, indexed or cited; that is the provider's decision. Turning off a training crawler does not remove a site from AI search answers. And some crawlers that fetch pages on a person's request say they may not follow these instructions at all.

llms.txt, with realistic expectations

Our platform can publish an llms.txt file: a short plain-text guide to a website's name, purpose, main pages and newest articles, written for AI systems. We generate it automatically from each site's settings and published content, and it updates as new posts are published.

We describe it accurately, though. It is a community proposal, not a standard that major AI providers have committed to using, and search engines say their AI features need no special files. It is a low-cost, optional extra. Crawlable pages, a sitemap and truthful structured data matter much more.

Instant change notices

When a page or post is published or changed, search engines normally learn about it on their next visit. Our platform can send instant notices through a protocol called IndexNow, which some search engines accept, so they can check the change sooner. Not every search engine uses it, and we say so in the settings rather than implying universal coverage.

What we do not do

Some tactics promise AI visibility and cross clear lines. We do not:

1. Create pages for every variation of a question to capture more queries. Search engines call that scaled content abuse.
2. Hide text for machines that visitors cannot see. That is cloaking.
3. Plant fake mentions or reviews so systems "notice" a brand.
4. Insert instructions aimed at AI systems inside pages, hoping to manipulate answers.
5. Promise citations or rankings. Nobody can honestly guarantee them.

These are not only against search engine policies. They also treat readers as targets rather than people to help.

Indexing waits for the right address

There is one more practical point. A website that is being rebuilt before it moves to its permanent address may deliberately keep some pages out of search engines until the move, so the old and new versions do not compete. Visibility work then focuses on everything that carries over automatically, such as canonical addresses that switch to the permanent domain once it is connected.

If you want the practical details, our posts on sitemaps and instant change notices and custom domains and canonical URLs explain the plumbing behind these ideas.

Frequently asked questions

Can I make ChatGPT or other AI assistants cite my website?

No one can guarantee citations. Clear, crawlable, genuinely useful pages with accurate structured data give AI systems the best chance to understand and use your content.

Does llms.txt improve search rankings?

There is no evidence it affects rankings. It is an optional guide for AI systems, and search engines say their AI features do not need it.

Should I block AI crawlers?

That depends on your goals. Search-oriented crawlers can help you appear in AI answers, while training crawlers collect material for model training. You can allow or block each group separately.