Most founders don’t sit down to write because they have spare time. They write because a strategy document, a launch post, or a positioning page has to say something that matters. The practical question is whether handing that job to an AI tool saves you time or quietly costs you the part that makes the document worth publishing.
The argument for AI is simple: it produces a draft in seconds, which is hard to ignore when you’re publishing on a schedule. The argument against it is less obvious but more consequential. Writing is not just a way to record what you already think. It’s how you find out what you actually think.
The Thinking Writing Forces You to Do
Paul Bakker, a software founder who writes about his own workflow, makes the case directly. He uses tools like Claude, Codex, and Gemini all day for coding, research, and digging through data. What he doesn’t use them for is writing, and his reasoning is worth sitting with: writing forces you to think deeply about the problem you’re solving and what you’re trying to communicate. That’s precisely why it’s hard.
When you write, you have to choose what comes first, what gets left out, and what deserves one sentence versus three. Those choices are the thinking. If you skip them by generating a draft from a few bullet points, you may end up with something that reads fine but doesn’t reflect the most important things you needed to say. You won’t know, because you never did the work that would have revealed the gap.
What You Lose When You Skip Straight to a Draft
The problem isn’t that AI tools are bad at writing. Bakker says the opposite: they’re good at writing. The issue is that they’re bad at thinking and coming up with new ideas. That distinction matters. An AI can organize your bullet points into clean paragraphs. It cannot know which of those points is actually the argument, which one you’re avoiding, or which one changes everything if you take it seriously.
This is where demanding publishing schedules become a trap rather than an excuse. If you’re publishing daily or weekly for SEO or AI search visibility, the temptation is to let the tool do more and more. But the parts of a post that earn trust — the specific example, the honest caveat, the reasoning that shows you understand the reader’s problem — don’t come from a prompt. They come from the founder’s own engagement with the material.
Why “Pretty Good” Is the Real Risk
AI-generated drafts rarely fail outright. They fail in a more dangerous way: they look good enough that you approve them with small edits. Bakker’s question is whether a document like that reflects the deep problem you’re solving. Probably not, he says, but you’ll never know because you skipped the step where you would have found out.
Founders often measure productivity by volume — how many posts went out this week, how fast the draft came back. That’s the wrong metric. Writing productivity is not about producing the most content in the shortest time. It’s about the ideas you bring and the impact those ideas have on your team, your customers, or the people reading. A short, well-thought-through document beats a long one full of fluff that never reaches the core problem.
Where AI Genuinely Helps a Founder Write
None of this means rejecting AI tools entirely. Bakker’s advice is explicitly not dogmatic. The point is to prevent AI from replacing your thinking, not to avoid it altogether. He uses AI after his initial writing and thinking are done. He asks it to read a draft and tell him what questions it raises, or what the single most important takeaway is. That’s like asking a colleague to proofread before you publish — useful, but completely different from asking someone to write the document from your bullet points.
A useful rule from his workflow: ask the AI questions, not for fixes. Don’t prompt it to rewrite or improve the document. Ask what’s unclear, what’s missing, what a sharp reader would challenge. Then you go back into thinking mode to decide how to address the feedback. AI can also help before you write — sifting through data, structuring information, finding patterns — so you understand the problem better before you sit down to think about solutions.
A Practical Order: Think First, Prompt Second
If you’re running a small business without a content team, the practical sequence looks like this: write the first version yourself, badly if necessary. Let the process force you to figure out what you actually believe. Then bring AI in as a questioner and editor. Then make the calls yourself. That order preserves the part of founder writing that creates value — the thinking — while still giving you a useful sounding board.
Tools like Serpio’s publishing workflow can handle the mechanics of keeping a blog moving, but the substance still has to come from the founder. If you want to see what that balance looks like in practice, a free plan is enough to test whether it fits your process without committing to a schedule you can’t sustain.
The honest question to ask yourself before your next post isn’t “Can AI write this?” It’s “If I don’t write the first draft, do I actually know what I’m trying to say?” How do you currently decide when to write for yourself and when to let AI take the first pass?
