Built to Be
Found by AI.
More people are asking ChatGPT, Claude, and Perplexity for recommendations instead of searching Google. This page explains what it actually takes to get named, and why most websites are invisible to AI without knowing it.
What "Optimized
for AI" Actually
Means
An AI names a business when it can reach the site, parse it accurately, and find clear factual answers worth quoting. That is the whole thing. It is not a label you add to a page. It is how the page is built.
This matters because the outcome is different from Google. In traditional search you compete for a slot in a list of ten links, and page two still exists. When someone asks an AI for a recommendation, they usually get two or three businesses named. There is no page two. You are either in the answer or you are invisible to that person.
The good news is that almost none of this is a trick. The same things that make a site fast, clean, and honest for a person are what make it readable to a machine. The bad news is that most sites fail at the very first step without realizing it.
Your Site May Be
Blocked Right Now
Cloudflare turns on AI crawler blocking by default for domains onboarded since mid-2025. Nobody chooses it. Nobody gets a notification.
It is applied at two independent layers. One is a firewall rule that refuses requests. The other is a managed robots.txt that injects a block on top of whatever the site owner wrote, overriding it entirely. Turning off one without the other fixes nothing.
What makes this so hard to catch: well-behaved crawlers read that file and leave on their own. They never knock. So no blocked requests appear in analytics. The site looks perfectly normal while being invisible to AI.
How to check it yourself, in ten seconds: open yourdomain.com/robots.txt in a private browser window. If it contains a block you did not write, listing names like GPTBot, ClaudeBot, or CCBot followed by Disallow, something is blocking AI access.
If you are on Cloudflare, the setting lives under AI Crawl Control, and both the Security tab and the Signals tab need checking. Both have to agree.
What Is Actually
Implemented
No jargon and no vague promises. This is what ships on every site I build, and you can verify every item in the code.
- Crawler Access, Verified Search and AI assistant crawlers are confirmed able to reach the site, at both layers, verified by loading the live robots.txt. This is the step most people skip, and without it everything else is wasted.
- Hand-Coded Semantic HTML Real headings, real paragraphs, real structure. None of the deeply nested markup that drag-and-drop builders generate. A language model parses a clean page accurately; with a bloated page it has to guess.
- Structured Data (JSON-LD) Business name, location, service area, price range, languages, and rating are published in a format a machine reads directly instead of inferring from page copy. The specific type for the trade is used, not a generic label.
- FAQ With FAQPage Schema Questions are written the way people actually ask them out loud, not the way a business would label them, and answers lead with the answer. The schema makes those answers machine-readable. This is the single most-cited piece.
- Content Signals in robots.txt Permission for search, for AI-generated answers, and for training is declared explicitly rather than left ambiguous. Most sites declare nothing, and some declare the opposite without knowing it.
- Specific, Verifiable Claims Real prices, real cities by name, real languages, real differences. An AI cannot repeat "quality service in your area." It can repeat "one-time $800, you own the code, no subscriptions."
- Performance and Clean Structure Fast pages, no layout shift, properly sized images, and a sensible heading structure. This helps people first. That it also makes a crawler's job easier is a bonus, not the point.
One Real Example
In July 2026, the owner of a real estate lending business in North Carolina asked ChatGPT to find her a web designer. OTM came up in the answer. She filled out the contact form that night and became a paying project.
It is worth noting what that was not. It was not a local search: she is more than a thousand miles from Kyle, Texas. It was not a referral from someone she knew. It was not paid advertising. An AI read the site, decided it was a legitimate answer to her question, and sent her.
A single case is not a guarantee and I am not presenting it as one. What it does show is that the channel is real right now, not a prediction about the future, and that a one-person business can show up in it with no ad budget.
When This Is Not
Worth Paying For
I do not sell AI optimization as a separate service, and I would be skeptical of anyone who does on a monthly retainer.
The reason is simple: almost all of this work is the same work that makes a site fast, well structured, and clearly written. A site built properly is already most of the way there. Charging separately for it would be charging twice for the same thing.
There are also cases where I am not the right fit. If your business runs on relationship-based contracts and does not take new customers from search, this will not move the needle. If you have a large e-commerce site or a web application, you need someone who specializes in that. And if anyone promises you the top spot in AI answers with a date attached, they are guessing.
If you have a site I did not build, a good portion of this can be fixed in place: crawler access, structured data, signals, and titles. What cannot always be fixed is the underlying markup from a drag-and-drop builder, and that is a rebuild rather than an adjustment. I will tell you which one you are looking at before you spend anything.
AI Search
Questions
It means an AI assistant can reach the site, parse it accurately, and find clear factual answers worth quoting. In practice that requires four things: the AI crawlers are not blocked, the page is clean semantic HTML rather than page-builder markup, key facts are published as structured data a machine can read directly, and the writing answers real questions in the words people actually use.
Load yourdomain.com/robots.txt in a private browser window. If it contains a block you did not write, listing crawlers such as GPTBot, ClaudeBot, or CCBot with Disallow, something is blocking AI access. On Cloudflare, check AI Crawl Control for allowed versus blocked request counts per bot. Then test directly: ask ChatGPT, Claude, or Perplexity a question your business should be the answer to and see whether you are named.
Cloudflare turns on AI crawler blocking by default for domains onboarded since mid-2025. It is applied at two independent layers: a managed firewall rule, and a managed robots.txt that injects a block and overrides what the owner wrote. Neither generates a notification. And because well-behaved crawlers read that file and leave voluntarily, the blocked requests never appear in analytics.
Yes, because it removes guesswork. A JSON-LD block states the business name, location, service area, price range, languages, and rating in a format a machine reads directly rather than inferring from page copy. FAQPage schema does the same for questions and answers. It is not a ranking trick; it is the difference between an AI having to interpret a page and simply being told.
They overlap heavily but optimize for different outcomes. Traditional SEO competes for a position in a list of links. AI search competes to be named as the answer, where there is no page two and often only two or three businesses mentioned. The same fundamentals help both: fast clean pages, accurate structured data, and honest specific writing. The difference is that AI rewards a clear direct answer near the top of a page far more than keyword placement.
Sometimes. Crawler access, structured data, robots.txt signals, and page titles can be fixed on most sites. What cannot always be fixed is the underlying markup: many drag-and-drop builder sites generate deeply nested code that is genuinely harder for both search engines and AI to parse, and that is a rebuild rather than an adjustment. An honest assessment comes first.
There is no fixed timeline and anyone promising one is guessing. Search-based AI can retrieve a page within days of it going live. Being known by a model without a search happens on much slower training cycles measured in months. Consistency of business information across the site, the Google Business Profile, and directories matters as much as the site itself, because AI cross-checks before naming a business.
For most local service businesses it is not worth a separate fee, and it should not be sold as an add-on. The work is largely the same work that makes a site fast, well structured, and clearly written. A site built properly is already most of the way there. Be skeptical of anyone charging a monthly retainer for AI optimization without being able to name the specific technical changes they make.
I'll check yours,
free.
Send me your domain and I'll tell you whether AI assistants can reach your site, what structured data you have, and what is missing. No obligation and no sales pitch. If everything is fine, I'll tell you that.