Your site may open on your phone while a search crawler cannot retrieve its public pages. Start with three checks: the crawler’s rules, the server response and the content returned. A missing ChatGPT mention alone does not establish a block.
Search access and training are separate choices
OpenAI’s crawler documentation distinguishes OAI-SearchBot for ChatGPT search from GPTBot for model training. You can allow search while declining training. ChatGPT-User handles certain user-initiated visits; it is not the search opt-out control, and robots.txt rules may not apply to those visits.
That distinction is the reason our home page uses OAI-SearchBot in its example. A rule targeting a different bot answers a different question.
Start with the file, then ask what it means
Open your own domain followed by /robots.txt. Save the URL, the date and the relevant lines before asking your provider to edit anything. The following is an illustrative block, not a finding about your business:
User-agent: OAI-SearchBot
Disallow: /
The provider should review the complete file, the exact public page and any more specific path rules. Avoid replacing the whole file with a snippet. Your store, account area or internal documents may need restrictions that should remain.
Ask for a plain answer: “Which rule applies to our services page, and what change are you proposing?” Keep a copy of the previous version so the change can be reversed.
Check the server too
A readable robots.txt does not prove that the next page request succeeds. Ask your provider to check whether the public page returns its content or a login screen, rate limit, access error or security challenge. Request evidence for the affected URL, rather than a general statement that the site is “AI ready.”
A request that merely names a crawler is a diagnostic clue. It does not prove that a real visit from that service’s verified network will receive the same response. Keep those two checks separate in the report.
Do not turn off the firewall to solve one blocked page. A targeted change should preserve account, payment and document protections.
Make the useful information readable
The next question is what an accessible public page actually says. Can a visitor identify the business, understand its services and find a working contact method? Are the essential details text, or only inside an image?
Google’s guidance for AI search features emphasizes ordinary search eligibility, crawl access and useful textual content. It requires no special AI text file and offers no inclusion guarantee. A custom file or a score cannot replace those basics.
For a Puerto Rico business, start with the information customers need to decide whether to contact you: service area, relevant services and how to reach the business. Publish only facts you can keep current.
Ask for evidence you can revisit
A useful handoff records the tested public URL, the applicable rule, the response observed, the change made and the remaining uncertainty. “Allowed,” “fetched” and “mentioned” describe different outcomes.
Try the free website analysis as a starting point. Then ask your provider to investigate the findings against the actual site configuration. If the work involves rebuilding the page, our guide to design and development explains the responsibilities involved. Review the website service and pricing when you are ready to discuss scope.
Nobody can guarantee a mention in ChatGPT. The practical goal is to remove an unintended obstacle and make the public information useful enough to evaluate.