Creator / Content Operator: the operator half is the job
ENGINEERING
SEPTEMBER 28, 2026

Creator / Content Operator: the operator half is the job

4 min read
BACK TO BLOG

Post four of ten. Making content got free. Being accountable for what it claims did not, and that gap is where the actual work moved.

This one is on the list as a single job, and it is really two. The split matters because AI did something dramatic to one half and nothing at all to the other.

The creator half got cheap

Drafting, editing, images, video, variations for five platforms. The marginal cost of producing a competent piece of content fell to roughly nothing, and it fell for everybody at the same moment.

That means it is not a differentiator. Whatever a model can make for you this morning, it made for several million other people this morning.

The operator half did not

Here is what happened to me, and it is the reason I take this seriously.

My website once said I had delivered over a hundred and fifty projects, that I had a satisfaction rate in the high nineties, and that I had worked with hundreds of businesses. None of it had happened. I am one engineer.

I did not write those sentences. A model generated the copy, I read it, it sounded like a website, and it went live. I found them weeks later, while auditing something unrelated.

Nothing in the production process was broken. The model did exactly what it was asked. Marketing copy contains confident figures — that is the genre. And when you read it back, an invented number does not announce itself. There is no visual difference between a figure you verified last month and one that was invented forty seconds ago.

What operating actually means

It means a register of every figure the copy is allowed to contain, each with a source precise enough to re-verify, and a checker with no model call in it that fails any draft containing a number not on the list.

Mine extracts every figure from the prose — money, percentages, counts of three or more digits, multipliers, written-out quantities like a third — and rejects anything unregistered. It also catches the opposite failure: figures that are real but marked internal and must not be published.

It does one more thing that I did not expect to need. The original copy was written in the first person plural: what the team had delivered, how many clients the company served. There is no team. So the checker flags we, our and us as well. Same class of problem as the invented project count — not a lie anybody chose to tell, just the default voice of the genre arriving unexamined.

Why this is the career

Because the constraint moved. It used to be can you produce enough. It is now can you stand behind what you produced, at a volume where you cannot personally remember writing any individual sentence.

The people who will do well at this are not the ones with the best prompts. They are the ones who built the boring apparatus that makes volume survivable: a source of truth, a check that runs before publication, and a rule about what happens when the check fails.

Volume without accountability is a liability that scales.

Where I come into this

Start smaller and free: the website check reads your live pages and reports what is on them, including the things a person scrolling past does not notice. A minute, no account.

If you publish figures your business depends on and nothing checks them before they ship, that is a register waiting to be built, and it is a weekend of work rather than a project. Tell me what you publish. If your review process already catches it, I will say so.

Post 4 of 10. The next one is a job I sell.

SHARE
4 min read
Read More

MORE ARTICLES

MSUFCU turned its own website off. That was the right call.
Engineering
Engineering

MSUFCU turned its own website off. That was the right call.

A PHP flaw became unauthenticated remote code execution with no workaround, so banks and governments went dark on purpose. What happened in Lansing, and why.

I gave an AI agent a wallet and a spending limit
Engineering
Engineering

I gave an AI agent a wallet and a spending limit

An autonomous client paid for its own API call in stablecoins. The payment was the easy half — the interesting half is the record of what it was refused.

The demo worked because you built it on your own data
Engineering
Engineering

The demo worked because you built it on your own data

The most expensive failure in applied AI is the one that looks like a success. Here is the test that catches it, and why it has to be able to fail.

How to read the arithmetic on a views guarantee
Engineering
Engineering

How to read the arithmetic on a views guarantee

A real offer promises 10 million views in 90 days or you do not pay. Here is the division that tells you what you are actually buying.