Every week the same promise arrives on LinkedIn, Instagram and YouTube. Install this MCP. Chain these two tools. Load this skill. Somebody else's twenty years of craft, gone by lunchtime, and yours to keep.
I sell one of the crafts on the list. So I checked.
What I actually did
For a few months I did what the videos said. The paid tools and the free ones. The MCPs, the skills, the workflows with the impressive diagrams. I pointed all of it at real client work rather than the toy tasks in the demos: positioning, content, search visibility, the odd site build.
I would like to report a disaster. There was none. The tools were good, some of them very good, and the plumbing between them often worked without complaint. The result was worse than a disaster. It was fine.
Fine is the problem
Fine means the output read fluently, was formatted properly, and could have been written by anyone about any company in the category. It was the average of everything the model had read, delivered at speed and dressed as a finished piece of work.
That last part is what makes it dangerous. A person who can judge the work sees a competent first draft with nothing in it and sends it back. A person who cannot judge the work sees a finished document and presses publish. That second person is the audience the tutorials are built for, and the result has a name now. Nobody making AI slop is lazy. Most of them followed the tutorial exactly.
The pipeline was never the issue. The brief was. A brief with no point of view comes back as work with no point of view, however clever the tools between the two ends.
The wrong audience
Here is the part the tutorials get backwards. An AI tool for marketing is useful to a marketer. A coding tool is useful to somebody who can read what it wrote. A design tool is useful to somebody who already knows why the layout is wrong. In each case the tool multiplies a judgment that was already there.
The tutorials are aimed at people with nothing to multiply. They promise a shortcut past the years in which you would have learned what good looks like, and they deliver a faster way to be wrong.
The two questions
Clients ask me, reasonably, why they should pay for judgment when the production is nearly free. I do not argue. I ask two questions.
You can translate your entire website into Japanese this afternoon. If you do not read Japanese, would you press publish?
You can draft a complete commercial contract before dinner. If you are not a lawyer, would you sign it?
Nobody has said yes to either. They already knew what they were paying for. It was never the hours. It was somebody prepared to look at the output and say, this is right, ship it, and to carry the consequences if it was not.
What is actually for sale
Professional work was never priced on production. It was priced on knowing what to make, what to leave out, and when the result is good enough to carry your name. AI has collapsed the cost of the first part and has not touched the second. If anything it has raised the price of judgment, because there is now far more plausible work in the world and somebody has to tell the difference.
So use the tools. I use them all day. Use them to do more of what you can already do well. Do not use them to impersonate a profession you have never practised, and be suspicious of anyone with a webcam who says you can.