All articles
AI SearchContent QualitySEO Practice

If AI Can Write a Hundred Pages Before Lunch, Why Do I Still Publish One Page at a Time?

Mehr Farooq·2026-11-14· 5 min read
A wooden desk overwhelmed by a tall stack of plain blank printed pages beside a laptop with a blurred unreadable screen, a half-finished cup of tea nearby in morning window light

What could the tools on my desk publish before lunch?

Let me start with the honest part, because this series falls apart without it: I use AI in my work every day. It drafts, it summarises, it checks my English, it argues with me about headings. If I pointed it at my UAE salary website tomorrow morning and simply approved everything it produced, that site could carry a hundred new pages before lunch — every profession in the Emirates, every guide, every comparison, all grammatically smooth and all wearing the same calm, confident tone. The pages would look finished. Some of their figures would also be invented — precise-sounding numbers with no source underneath them, produced because a salary page is expected to contain a number and the machine obligingly supplied one. Nobody reading would know which figures were real. That is not a writing problem. That is exactly the problem.

The technology, then, is not the obstacle. It has not been the obstacle for two years. So when people ask why my sites grow the way they do — a small batch of profession pages at a time, one post a day on this blog — the honest answer is that the slowness is a decision. This post is me explaining that decision to myself as much as to anyone else.

Read next
If an AI Answers Your Customer Before Your Website Does, What Is Your Website Still For?
Open

What did five years of other people's pages teach me about volume?

Before I published a single page of my own, I spent five years at the unglamorous end of this trade: backlinks, profiles, forum posts, map listings — page after page produced for other people's websites. I watched an entire economy of fast content from the inside. Thousands of pages written to exist rather than to be read, spun and respun, published by the hundred — and then, one search update at a time, buried by the hundred. Nothing about that era failed because the writing was slow. It failed because nobody involved could have answered a simple question about any given page: is this true, and how do you know?

When the machines that judge pages started asking that question at scale, volume stopped being an asset and became evidence. I did not learn that lesson from a course. I invoiced it, monthly, for five years — and I am not going to unlearn it now just because the fast pages have become better dressed.

What do I still refuse to hand over to the machine?

Three things, and they are precisely the three things that make my working day slower than it needs to be. First, the figures. Every salary range on my UAE site is traced to a named source before it is published, and when the sources disagree — which they do, constantly — the disagreement goes on the page instead of being quietly averaged away. A figure I cannot trace does not get softened or rounded into print; it gets deleted at the desk, however useful it would have looked.

Second, the judgement calls: which pages deserve to exist at all, which angle is the honest one, which claim I am willing to stand behind with my name at the top of the site. Third, the last read. A machine can produce a page that sounds right. Only the person whose name is on the door can decide whether it is right — and that decision cannot be delegated without the name becoming decoration.

Why does one careful page outlast a hundred fast ones?

Because the web's problem has quietly changed sides. For twenty years the scarce thing was content — whoever published the most, won the shelf. Now content is close to the cheapest substance on earth, and the scarce thing is a page a reader, or a machine, can rely on without checking it twice. My salary site did not reach a Domain Rating of 31 by being the biggest site in its niche; it is not, and it will not be. It got there profession by profession, because other websites — forty-six of them at the last count I trust — found its numbers worth pointing to. And numbers are only worth pointing to when somebody gathered them carefully enough to defend them in public.

There is a second, quieter reason. The AI answers everyone now complains about are assembled out of pages — somebody's pages. A page built slowly, from named sources, with its working shown, is the kind of page an answer gets built from. A hundred fast pages are the kind that get averaged in and forgotten. Or, on a bad day, quoted wrongly at scale, with your name still attached to the error. Speed decides how much of that risk you publish before lunch.

A woman's hands with a pen checking figures in a plain paper notebook beside a simple calculator and reading glasses on a worn desk in soft daylight

If AI can write a hundred pages before lunch, why do I still publish one page at a time?

Because publishing was never the bottleneck. Being believed is. The tools on my desk have made the writing nearly free, and I am grateful for them every working day — but gratitude is not a handover. The figures still get traced. The disagreements still get printed. The page still gets its last read from the person who will answer for it. My sites are small, and they will stay slower than they could be, and I have made my peace with that trade: one page at a time, each one able to survive the question the fast pages never could — is this true, and how do you know?

So here is today's question: if the tools on your desk could fill your website by lunchtime tomorrow, which parts of the work would you still refuse to hand over — and is your honest answer a principle you chose, or just a habit you have not examined yet?

A single blank sheet of paper placed in the middle of an otherwise empty wooden desk, a fountain pen resting beside it and a small potted plant at the corner in calm window light

Want SEO built on proof, not promises?

I build and rank websites for home service businesses — and I test everything on my own sites first.