I set up structured data, created an llms.txt file, and let AI crawlers like ChatGPT and Perplexity read my site. I automated a monthly routine to keep it current, because consistency matters more than one-time setup. Sites I maintain now appear in AI answers.
When someone searches for an answer in ChatGPT or Perplexity instead of Google, that AI tool cites the websites it reads from. Those citations are traffic. But most websites are set up only for Google. They are invisible to the AI answer engines that are now real sources of visitors.
People sometimes call this task generative engine optimisation, and most of the advice on it is guesswork. I set up my own website for these tools and have done the same work for clients. I have seen sites I work on cited in AI answers. What surprised me was not that it works, but how much of it is just showing up consistently and not how clever you are on any single day.
What the setup actually is
On my site deepconnects.com, I did three things. I added structured data, which is markup that tells machines what the page is about, what problem it solves, and what words matter. I created an llms.txt file that sits on the root of the domain and describes what the site covers in plain words. And I made sure my robots.txt does not block AI crawlers like ChatGPT, Claude and Perplexity. I let them in instead.
On a client site called BabyJunctions, I went further. I automated the whole SEO routine, not just the setup. That meant automatic internal linking based on topic relevance, a monthly pass that checks whether the structured data is stale, and a process that keeps the site fresh.
The technical part is small. The discipline is doing it over time.
Why once is not enough
When I first automated Instagram posting and ran an AI music channel on YouTube, I expected the best individual posts to drive everything. What I actually saw was that consistency mattered more than quality. An account that posts on schedule builds an audience that expects it. An account that puts out one brilliant piece and then goes quiet starts over every time.
It is exactly the same with AI answer engine visibility. You can set up perfect structured data and an llms.txt file, but if the site goes stale, if the pages do not stay current, if there is no process to check that the markup is still accurate, then the AI tools move on. They cite the site that has the current answer, not the site that had the perfect answer once.
What I see happening now
I have seen sites I work on cited in AI-generated answers. Not always, and not for every question, but when someone asks about a specific problem these sites solve, they show up. The difference between those sites and the ones that do not get cited is not that their content is more clever. It is that they have someone accountable for keeping it current, and there is a routine that actually runs.
This is where automation earns its place. Not in making each page brilliantly written. In making sure the foundations stay solid. The structured data stays true. The llms.txt stays accurate. The internal links point to the right current page. That is what lets AI tools trust your site enough to cite it.
What to actually check
If you are thinking about whether to set up your site for AI answer engines, here is what I would check first.
- Look at your robots.txt. If you are blocking ChatGPT-4, Claude, or Perplexity crawlers because you thought they were scrapers, you are invisible to those tools on purpose. Most sites that block them are not doing it consciously. It is a leftover setting from when people were more afraid of being scraped. You can let them in. They cite you as a source.
- Second, audit whether your key pages have structured data. Not everywhere. But the pages that explain what you do, what problem you solve, the case studies that show what you have built. Schema markup that clearly says 'this is a service offering' or 'this is a case study' makes you findable and quotable. It is not hard to add. It is hard to remember that every page needs it.
- Third, put together a process, however small. Even just a monthly check: do the main pages still have current information? Do the internal links still point to the right destination? If you have a services page, does the list match what you actually offer? This is the difference between a site that AI tools cite once and a site they cite repeatedly.
None of this is technically complicated. The reason most sites are invisible to AI answer engines is not technical complexity. It is that they were built for Google, and letting AI crawlers in, adding metadata, and keeping a process going takes separate thought.
Most site owners do not think about it until they see a competitor cited in an AI answer and wonder why they are not. By then, the AI tools have been training their models without knowing what you do.
The real work is not the setup. It is keeping it current.
