Reading time: about 11 minutes
This is the scariest case and the one that deserves it least. You are not going to learn to code: you are going to ask the AI for a file, drag it onto a site that publishes it, and have an internet address. The hardest part is deciding what goes on it, which is precisely the only thing the model cannot do.
Step by step
-
Write on paper what it has to say
Before opening anything. What a pharmacy page needs is short and always the same: name, what you are, services, hours, address with a Maps link, phone and how to get there.
What it does not need: a blog, a shop, a contact form (which also obliges you to a privacy policy and to answering it) and an "our philosophy" section.
If you have this on paper, the rest is mechanical. If you do not, no tool will solve it for you.
-
Ask for the whole thing, in one single file
This is the only technical thing here and it comes explained: a web page is a text file that the browser knows how to draw. It is called HTML. If you ask for "one HTML file with everything inside", you get a single file that works on its own, with nothing to install.
The key instruction in the prompt is "everything in one single file". Without it you get three separate files and have to arrange them in folders, which is exactly where people give up.
Build me a pharmacy web page. ONE SINGLE HTML FILE, with the styles inside the file itself (a <style> tag). No libraries, no links to anything external, no JavaScript. DETAILS: - Name: Pharmacy ... - What we are: a neighbourhood pharmacy in ..., since ... - Services: list yours - Hours: yours, Saturdays included - Address: street, number, town - Phone: ... - Google Maps link: paste your listing’s short link HOW IT SHOULD BE: - Perfectly readable on a PHONE. That is where nearly everybody will see it. - Hours and phone at the VERY TOP and large. - The phone number tappable to call. - Sober colours. Green and white. No gradients, no animations. - Large text: older people will be reading it. - Include the page title and a description for Google. WHAT IT MUST NOT HAVE: - A contact form. - Cookies, analytics or anything loading from outside. - No images (I will add those later). Give me the complete file, in full, from the first line to the last.
-
Save it and open it on your own computer
Copy everything it gave you and paste it into Notepad (Windows) or TextEdit in plain-text mode (Mac). Save it as
index.html. The name matters: it has to be exactly that.Beware of Word. Paste it into Word, save, and it adds invisible things that stop the file working. Notepad or nothing.
Now double-click the file. It opens in your browser and you are looking at your page. It is not on the internet yet — only you can see it — but it exists, and seeing that is what takes the fear out of the two remaining steps.
-
Read it as a customer would and fix what is wrong
Before publishing. Check the hours digit by digit: that is what most often gets copied wrong. Tap the phone number. Tap the Maps link and check it takes you to your pharmacy and not another one.
And narrow it: drag the browser window edge until it is phone-width. If anything overflows or needs sideways scrolling, say exactly that: "on a narrow screen the hours table overflows".
To change anything, do not touch the code: say "change Saturday hours to 9:30-14:00 and give me the whole file again". Always ask for the whole file, not the fragment: pasting a fragment in the wrong place is the one way to break it.
-
Publish it by dragging the file
There are services that publish a page by dragging the file onto them, free and with no card. Search for "publish html free drag and drop" and pick one; which one does not matter, the procedure is the same: you drag, you wait twenty seconds, you get an address.
That address is ugly — it has the service’s name inside it — and it is good enough to start. You can put it on your Google listing, on your receipts and in the window right away.
If later you want your own address, a domain costs about ten euros a year and is pointed at that same page. But that comes after: publishing with the ugly address today is worth more than the pretty address next month.
-
What DOES carry legal consequences, and why this page has none of it
This case’s page is deliberately built so that you do not have to do anything else. No form, no cookies, no analytics, nothing loading from outside: which is why it collects no data and creates no new obligations for you.
The moment you add a contact form, a chat widget or analytics, you are in different territory: you need a legal notice, a privacy policy and — depending on what you add — a cookie notice. A prompt does not solve that, and AI-generated legal text for your specific business is exactly the kind of thing that looks fine and is not.
And one thing is required from day one even though the page collects nothing: as the holder of a business you must be identifiable. Trading name, tax number and a contact. Put those three at the foot and, for anything beyond that, ask your adviser — not a chat.
A full example, and what is wrong with it
The step 2 prompt with a neighbourhood pharmacy’s details. I am not pasting the whole file — it is 120 lines — but the four places to look.
<link href="https://fonts.googleapis.com/css2?family=Poppins..." rel="stylesheet">
<h1>Example Pharmacy</h1><p>Your trusted neighbourhood pharmacy</p>
<p>Hours: Mon-Fri 9:00-14:00 / 17:00-20:30 · Sat 9:30-14:00</p>
<a href="tel:+34900000000">900 000 000</a>
"Here is your complete page. Just save it as index.html and open it in your browser."
And now, what is wrong with that reply:
- The first line loads a Google font. The prompt said "no links to anything external". And there it is, because it is what 95% of the pages the model has read carry. Search the file for
http: the only thing that should appear is your Maps link. If there is more, ask for it to be removed — and with that the page stops loading anything from outside and creates no new obligations for you. - The phone is in international format and the visible one is not. The tappable format is right, but check the number digit by digit. It is the single most often miscopied piece of data on the whole page, and a phone number with one digit wrong is worse than no phone number: people ring and nobody answers.
- "Your trusted neighbourhood pharmacy" says nothing. It is the same sentence as the Google listing and half the counters in the country. On a one-screen page that line occupies the space where what you do should go: naming three services in the same space informs and is also what gets searched.
- And it put the opening hours on a single line of text. It fits on a desktop and on a phone it breaks in the wrong place, leaving a stray fragment on its own line. It is the commonest layout failure in this case and you only see it by narrowing the window — which is why step 4 says to. Ask for the hours as a list or a small table.
Four things and none needs touching the code: you say them in one sentence and it returns the whole file corrected. All you have to do is know what to look for, which is what this case is about.
And the check that covers all four: search for http, tap the phone, tap the Maps link, narrow the window. Two minutes, and it is what separates a page that works from one that merely exists.
And if your pharmacy is not like that
When it does not work first time
index.html.txt. The operating system hides the extension, so it looks right. In Notepad, when saving, choose "All files" in the type dropdown and type the full name with .html. It is the number one stumble here and has nothing to do with AI.</html> — or it was pasted through a word processor and mangled. Ask for the whole file again and paste it into a plain notepad, never into a word processor.http: the only thing that should appear is your Maps link. If there is more, ask it to remove it.form. And if it appears, remove it without hesitating: a form collects data, which obliges you to a privacy policy and to answering it — exactly what this case is designed to avoid.