AI-Directed, Not AI-Generated
"We use AI" has stopped meaning anything. Every agency says it, every tool claims it, and the output all looks the same. The distinction that actually matters is who is doing the deciding.
- AI-generated means the tool picks the topic, the angle and the words. AI-directed means a person does, and the tool executes.
- The difference shows up in whether your content could have been published by a competitor with the same prompt.
- Every piece 9xNine publishes is reviewed by a human operator before it goes out. That is the product, not a disclaimer.
Give ten businesses in the same industry the same AI tool and the same brief, and you will get ten near-identical libraries of content. Not because the tool is bad, but because nothing in that process is specific to any of them. The tool has no idea which of the ten is actually the best at anything.
That is the whole problem with AI-generated marketing. It is competent, fast, cheap, and completely interchangeable. It fills a calendar without moving a business.
What the difference actually looks like
- The tool picks the topic from a keyword list
- The angle is whatever the model defaults to
- No knowledge of what makes you different
- Volume is the only available lever
- Nobody reads it before it publishes
- A person picks the topic against a positioning brief
- The angle is chosen, and defended
- Built on a documented, specific edge
- Judgement is the lever; volume follows
- A human operator reviews every piece
Read your last published piece and ask whether a direct competitor could have published it word for word. If the answer is yes, the tool wrote it — and the tool does not know you exist.
Where the human actually sits
"Human in the loop" is usually a euphemism for a spellcheck. Here is where the decisions genuinely get made in an AI-directed process:
-
Before anything is written
A person decides what this business is actually the obvious choice for, and writes it down. Everything downstream argues that one case from a new angle.
-
At the brief
The topic, the angle and the claim are chosen by a person against that positioning — not pulled from a keyword volume report.
-
At the draft
AI does the drafting. That part is genuinely faster, and pretending otherwise would be dishonest. What comes back is a draft, not a deliverable.
-
Before it publishes
A human operator reads it end to end: is the claim true, is it ours, would we say it out loud to a client. Nothing publishes without that pass.
AI-directed is not slower than AI-generated in any way you would notice, but it is not free either. The cost is judgement, and judgement does not scale infinitely. Anyone promising unlimited output and full human review is telling you one of those two things is not real.
How it compares to the two obvious alternatives
- Versus a DIY AI tool: the tool gives you a draft. It cannot tell you whether the draft was worth writing.
- Versus a traditional agency: same judgement, without paying junior-writer hours for the drafting step that AI genuinely does well.
- Versus doing nothing: the archive you did not build this year is the one you will wish you had next year.
The tool is not the differentiator. The judgement behind it is.
Does a human really read every piece?
Yes. That is the line we hold, and it is the reason output is scoped rather than unlimited. If we ever could not hold it, we would change the offer rather than quietly drop the review.
Will my content sound like AI wrote it?
It should not, and that is a reviewable thing rather than a promise. Read a piece before it publishes and tell us if it does — that feedback goes straight back into the positioning brief.
Can I just do this myself with the same tools?
You can, and some owners should. The tools are not the hard part. The hard part is deciding what you are the obvious choice for and then holding a publishing cadence for twelve months.
Find your edge
A short, outside-in read on what already makes you the obvious choice.
