Insights / Visibility & Authority
AI, a writer or both? How to choose the right content workflow
The real choice isn’t tool versus person. It’s who has the expertise, who reviews it, how much you need and what happens if it’s wrong.
Most advice on this question sets it up as a contest. Hire a writer, or use AI. Pay more for quality, or pay less for speed.
That framing is out of date. Most businesses producing decent content now use some mix of both. The useful question is which mix, for which kind of content.
The answer depends less on the tool than on four things: who has the expertise, who reviews the work, how much you need, and what it costs you if something is wrong.
The real question isn’t “AI or a writer?” It’s “where does the expertise come from, and who is accountable for getting it right?”
Three workflows, not two options
AI-led. Someone on your team prompts an AI tool, edits the result and publishes it. Fast, low direct cost. Works when the subject is simple and the person editing knows it well.
Writer-led. A writer researches, interviews and drafts, possibly using AI along the way. Slower, higher direct cost. Works when the subject needs explaining, structuring or a voice your team doesn’t have time to find.
Hybrid. An expert supplies the substance through an interview or notes. A writer or editor structures it, with AI helping on transcription, outlining and tightening. The expert approves it. It’s the method described in getting content out of your head.
None of these is right for everything.
How they compare
| Factor | AI-led | Writer-led | Hybrid |
|---|---|---|---|
| Expertise access | Only as good as the prompt and editor | Depends on the writer’s research and access to your people | Comes directly from your expert |
| Review | Every fact needs checking; easy to underestimate | Writer checks sources; expert still reviews | Expert reviews for accuracy; editor for clarity |
| Volume | High, if review keeps up | Limited by writer time | Moderate; limited by expert time |
| Risk if wrong | Higher on technical, regulated or advisory topics | Lower, if the writer has access to experts | Lowest, when the expert signs off |
| Accountability | Whoever pressed publish | Writer for the draft, you for the claims | Named expert approves; editor owns process |
A starting point, not a rule. Your team and topics will shift the balance.
Count the review, not just the draft
AI drafts are cheap to produce. They aren’t always cheap to publish.
AI tools can state things confidently that aren’t true, mix up details, or produce something accurate but generic. Somebody has to catch that. If the only person who can catch it is your most senior expert, the real cost of a “free” draft might be an hour of their time.
A writer’s draft has its own review cost, especially if the writer doesn’t know your field. The difference is where the work sits: research before the draft, or correction after it.
When you compare options, add up the whole thing:
- Time to gather the expertise.
- Time to draft.
- Time to fact-check and correct.
- Time for the expert to approve.
- Time to update it when things change.
A fast draft that needs an hour of fixing isn’t fast. It’s just fast to start.
Choosing the right mix
Sort content by stakes.
A social post announcing office hours is low risk. A page explaining tax treatment, a medical service or a safety requirement is not. The higher the cost of being wrong, the more the expert needs to be involved.
Check where the expertise lives.
If the person writing knows the subject, AI-led drafting with careful review can work. If the knowledge is in someone else’s head, you need an interview first, whoever drafts.
Be realistic about volume.
Producing more is only useful if each piece answers a real question. Ten pages that say the same as everyone else won’t help a buyer choose you. See the standard we hold AI-assisted content to.
Name who approves.
Every piece of content needs one person who confirms it’s accurate. Not the tool. Not “the team”. A name.
Decide where AI helps.
Transcription, summaries, outlines, headline options, readability edits and first-pass drafts from good source material are sensible uses. Facts, opinions and examples should come from people who know.
A hypothetical split
Say you run a 15-person physiotherapy clinic. Weekly social posts about class schedules: AI-led, reviewed by the front desk. Service pages explaining treatment approaches: hybrid, with a physiotherapist interviewed and approving every claim. A guide for referring physicians: writer-led from an interview, reviewed by the clinical lead before anything goes out.
Three kinds of content, three workflows, one rule: the higher the stakes, the closer the expert stays.
AI or a writer is the wrong argument.
Start with who knows the answer.
Then decide who helps them say it.