Should a founder use AI to write the blog posts, memos, and emails that carry their name? The short answer: use it to start, not to finish. AI is a useful brainstorming partner and first-draft accelerant—but it should rarely produce the final prose your readers consume. Here’s why, and a practical workflow that keeps AI in its proper place.
Writing Is Thinking—and AI Can’t Do That Part for You
The deepest argument against letting AI write for you is simple: writing is a core part of thinking, not a packaging step after thinking is done. When you write about a topic, even one you know well, the act of putting ideas into words exposes gaps, reveals assumptions, and often produces genuinely new ideas. As essayist Paul Graham has observed, half the ideas that end up in a finished piece are ones you did not have before you started writing it. The writing is the thinking.
For a founder, this matters more than for most people. Your blog posts, product updates, and positioning documents shape how customers understand your business. If you outsource that cognitive work to a model, you skip the part where you might notice that your pricing logic is shaky, your differentiation is unclear, or your customer’s problem is different from what you assumed. Clara Collier, writing on Patrick McKenzie’s podcast, put it plainly: in substantive writing, there is no part of the process where you are not thinking and changing your mind—from the outline to individual sentences.
This has direct implications for your content strategy. A seo content strategy built on fully AI-written posts may publish more often, but each article communicates less of the judgement that makes your expertise worth hiring. When the model does the final thinking, the reader gets generic competence instead of your actual point of view.
The Hidden Costs of AI-Proposed Prose: Vagueness, Wrongness, and Misleading Readers
The second problem is more concrete. AI writing is vague and wrong in hard-to-notice ways. Modern models produce fluent, plausible text that reads well on a first pass—and that’s exactly the danger. Errors hide inside confident phrasing, and vague sentences sound like insight until you try to act on them.
The third problem is relational: writing with AI, and not labeling it as such, is rude and misleading. Your readers assume the words on your blog or in your email represent your thinking. If a model wrote them, and you imply otherwise, you are letting someone spend their attention and trust on a synthetic artefact dressed up as judgement. That’s not a trivial harm—it’s a slow drain on exactly the trust small businesses depend on.
None of this means AI has no place in writing. It means AI’s place is earlier in the process.
Where AI Actually Belongs in Your Writing Workflow
A useful mental line separates generation and editing from substantive prose. AI makes sense for:
- Brainstorming topics and angles you might not have considered.
- Outlining a piece before you write it.
- Producing rough drafts that you will substantially revise.
- Searching for information and organising research.
- Giving feedback on a draft you wrote yourself.
It also works for line and copy editing: tightening sentences, fixing pacing, and catching forced phrasing. But even then, every edit must be deliberately accepted or rejected by you, not silently inherited from the model. The moment you let the tool shape the substance—not just the expression—you lose the thinking benefit and inherit the vagueness risk.
If you’re building a daily publishing habit as a busy founder, tools like Serpio’s article workflow can handle the mechanical parts—topic discovery, draft generation, formatting for AI search—while leaving final approval and the substantive voice in your hands. That division of labour respects the line: machines propose text; humans decide what ships.
A Practical Founder’s Workflow: Draft Fast, Edit Slow
Here’s a repeatable process that keeps AI fast without letting it finish the job:
- Write the messy first draft yourself—or at least the outline and key arguments. This is the thinking phase, and it cannot be delegated without cost.
- Use AI to generate a parallel draft from your outline, mainly to surface angles, counterexamples, or phrasings you may have missed.
- Compare the two. Where does the AI version sound confident but vague? Those are the spots to strengthen with your own specifics.
- Rewrite the final version in your own words. Accept specific line edits from AI only after reviewing them one by one.
- Publish with your voice intact. If you used AI for brainstorming or a rough draft, you do not need to label it—the ideas and final prose are yours.
This does not add much time; in many cases, brainstorming and a generated parallel draft make you faster than starting from a blank page. What it restores is the part only you can do: deciding what you actually think, and saying it clearly.
When Delegating More Might Make Sense
There is an honest counterpoint to be made. AI writing is less effortful and much faster than writing yourself. For a time-pressed founder shipping product, handling customers, and trying to keep a blog alive, that speed is genuinely valuable. And future AI models may become good enough that delegating the whole research-and-writing pipeline becomes sensible—at which point delegating just the writing half may not even be the right split.
But for the models that exist now, and the ones likely to exist in the near future, the tradeoff still favours you doing the substantive thinking. The question a founder should ask is not “Can AI write this faster?” but “What am I giving up when it does?”
Counterpoint: Speed and Low Effort Are Real Advantages
Let’s state the opposing view fairly. If your content strategy is purely about volume—getting functional posts live to support keywords and internal links—then an AI-written draft may feel like the only sustainable option when there is no content team. The quality bar may be low enough that vagueness does not obviously cost you traffic today.
The risk is what that choice trains into your publishing habit. Each fully delegated post is a thinking opportunity you did not take. Over months, that compounds into content that ranks but does not persuade—posts that answer a question without ever demonstrating why you are the one worth hiring. A seo content strategy that optimises for clicking and skimming may technically grow organic reach while quietly undermining the reason that reach should convert.
Serpio’s approach reflects this tension: the platform can generate timely drafts and structure them for AI search, but the workflow keeps review and editing in front of you. Plans for daily publishing exist for founders who want the volume without pretending the final voice is fully automated.
Still, the fact that a tool can write for you does not mean you should let it write for readers under your name. The distinction between “AI helped me draft this” and “AI wrote this” is not just semantics—it is the difference between a useful assistant and a misleading substitute.
What has your experience been? Have you found a line where AI genuinely helps your writing without flattening it—or do you find it creeps further into the process than you intended? I’d like to hear where you draw the line.