The Invisible Storefront
This week I ran a technical scan for a specialist retailer with fifty years of trading history. The website is good. Prices are clear, the product pages are rich, and customers find what they need. To a human visitor, nothing is wrong.
To an AI assistant, the shop does not exist.
When ChatGPT, Claude or Perplexity tried to read the site, every single page returned the same answer: access denied. Not one product, not one price, not even the contact page. A business that has served customers for fifty years is closed to the fastest-growing referral channel in retail, and nobody in the company decided that. Nobody was even asked.
How this happens
A growing share of shopping journeys now starts with a question to an AI assistant. Which turntable should I buy? Who installs heat pumps near me? Which supplier ships to Ireland? The assistant answers by reading websites. If it cannot read yours, you are not in the answer.
The reason is uncomfortable. In 2025, the security services that sit in front of most commercial websites began blocking AI crawlers by default. Hosting companies switched the same blocks on across thousands of client sites at once, mostly to save server load: AI crawlers visit constantly and in volume, and blocking them cuts traffic and cost for whoever runs the servers. These decisions were reasonable from an infrastructure point of view. But they were made at the infrastructure layer, by default settings and busy administrators, and almost never by the person who owns the revenue.
There is a second layer to it. In months of analyzing companies of every size, I keep finding the same thing: awareness of agentic visibility barely exists inside marketing teams. This is a new area of marketing. Most teams do not yet have the knowledge base to spot the problem, let alone the technical vocabulary to challenge their agencies on it, and most agencies are not looking for it either. So the block sits there, undetected on both sides of the client and agency relationship.
The result: a meaningful share of businesses are invisible to AI assistants today and have no idea. Your website analytics will not show it. Your search rankings will not show it. The block is silent.
Check your own site in sixty seconds
You do not need to be technical. You will use a small built-in program that asks your website for its homepage twice: once pretending to be a normal browser, and once identifying itself the way an AI crawler does. If the two answers differ, you have your diagnosis.
On a Mac: press Cmd and the space bar together, type Terminal, and press Enter. A plain text window opens.
On Windows: press the Start button, type PowerShell, and press Enter.
Now copy the five lines below, paste them into that window, and press Enter. One change first: replace YOURSITE.com with your own web address, keeping the rest exactly as it is.
curl -s -o /dev/null -w "as a normal browser: %{http_code}\n" -A "Mozilla/5.0" https://YOURSITE.com
curl -s -o /dev/null -w "as ChatGPT's crawler: %{http_code}\n" -A "GPTBot" https://YOURSITE.com
curl -s -o /dev/null -w "as ChatGPT answering a question: %{http_code}\n" -A "ChatGPT-User" https://YOURSITE.com
curl -s -o /dev/null -w "as Claude's crawler: %{http_code}\n" -A "ClaudeBot" https://YOURSITE.com
curl -s -o /dev/null -w "as Perplexity's crawler: %{http_code}\n" -A "PerplexityBot" https://YOURSITE.com
(On Windows, if you see an error on the first line, type curl.exe instead of curl at the start of each line. Everything else stays the same.)
If you are not technically minded and run into any trouble, asking your AI assistant for help is probably the easiest solution.
Reading the result. The number 200 means the door opened. The number 403 means the door was slammed. If your browser line says 200 and the crawler lines say 403, your website is currently invisible to AI assistants. That was the retailer's exact result, on every page.
One honest caveat: this test is a strong signal, not a court verdict. Some security setups treat a test from your laptop differently than the real crawler. So if you see 403, do one more thing: ask whoever manages your website to open your security dashboard and check the bot settings. In Cloudflare, the most common service, look at the bot settings under Security and at the AI Crawl Control section. The dashboard is the ground truth.
If you are blocked
Do not panic. In most cases nobody did anything wrong; a default did it. But do treat the discovery seriously, because what you have found is not a technical fault. It is an unowned decision.
Whether machines can read your store is now a commercial choice with revenue attached, the same class of decision as which marketplaces you list on or which countries you ship to. It deserves an owner, a deliberate yes or no, and a periodic check that the setting still matches the intention. Some businesses will decide to stay closed to AI crawlers, and for some that will be right. The failure mode is not the block. The failure mode is not knowing.
Fifty years of reputation can be absent from the shelf where a growing number of customers now look. It takes sixty seconds to find out if that is you.
Who in your business owns that decision today?
(c) 2026 Paul F. Accornero / The AI Praxis™. All frameworks and terminology proprietary: paulaccornero.com/legal-and-ip