The draft is the cheap part.
The question is not whether a machine helped write it. It is who read it afterwards, what they were allowed to reject, and whether anyone checked the facts it states so confidently.
Good at scaffolding. Bad at being sure.
What a draft is good for
- Getting to a complete structure fast, so the argument is visible before anyone polishes a sentence.
- Covering the obvious ground — definitions, comparisons, the parts a reader expects and would notice missing.
- Volume. Forty articles in a cluster is a scheduling problem, not a creative one.
- Consistency of format across a set, which is tedious for a person and trivial for a tool.
Where it goes wrong
- Confident wrong facts, stated in the same voice as the correct ones. This is the failure that costs you something.
- Citations that do not support the sentence they are attached to, or do not exist.
- Averaged opinions. Asked what to think, a model returns the middle of what has been written, which is the one position nobody remembers.
- House style. It does not know your product's name for the thing, or which claim your legal team struck last year.
A writer drafts. An editor decides.
Every article passes a second person before it reaches you, and that person can send it back. The full path is on how it works.
- 01
The brief is fixed before anyone writes: topic, keywords, length, link policy, whether references are required.
- 02
A writer drafts with whatever tools they use, and is responsible for what the draft says — not for having typed it.
- 03
An editor reads it against that brief. Facts, citations, structure, house rules. They approve it or send it back with a reason.
- 04
You see it at approval, never before. If it is wrong, it comes back to the same loop rather than to you.
The four questions everyone asks.
- Do you use AI to write the articles?
- Our writers do, and they choose their own tools. We do not run a generator and call the output a service — the platform is the brief, the workflow and the review. What we commit to is what happens after the draft exists.
- Will a detector say this was written by AI?
- We do not know, and neither does anyone selling you a number. Detection tools disagree with each other and misclassify human writing routinely. We make no claim about what any of them will report, and you should be wary of a supplier who does.
- So what is the actual difference?
- An editor reads every article against the brief before you see it, and can send it back. That is a person with the ability to reject the work, which is the part a prompt cannot supply.
- Is AI-written content bad for search?
- Search engines say they reward helpful content regardless of how it was produced, and penalise thin content produced at scale to game rankings. Both halves of that are worth taking at face value. The variable you control is whether the piece is worth reading.
Elsewhere in this section
Strategy overview
How we plan a content programme, and the pieces it is made of.
Free GEO audit
Answer five questions and get a written content audit — gaps, clusters, competitors and thought-leadership angles.
GEO & SEO resources
How answer engines pick what to cite, and the search fundamentals underneath.
Thought leadership
What separates a founder's argument from a summary of the category.
Read three and judge for yourself.
Three complete articles from one order, shown with the brief each was written from — nothing trimmed.