A founder I spoke with this spring had been writing the about page for her firm for fourteen weeks. She is articulate, direct, knows the business inside out. She had also opened the document seventeen times and produced four hundred words across the lot of them. The blank page was winning. When I asked her what she was stuck on, the answer was not the content. She knew what she wanted to say. She just could not bring herself to start saying it, in writing, alone, on a Tuesday morning, when the children were back at school and the cursor was blinking at her.
The blank page is the most expensive piece of paper in your week. The actual writing rarely takes more than an hour. The fourteen weeks of not writing it cost the firm a website that could not introduce its founder, a sales call that opened with apologies for the site being out of date, and the slow erosion of the founder’s own confidence that she could write at all. AI removes the blank page reliably, in seconds. That is the move worth understanding. The first draft is rarely the final draft, but the first draft was never the bottleneck. The staring at white was.
What is a first pass with AI?
A first pass is a draft a model generates from your brief, used as raw material rather than a finished product. The output is rough, often off-angle, sometimes factually wrong. It is also on the page, which changes the cognitive task entirely. You stop creating from nothing and start reacting to a draft, and reacting is much smaller work than producing.
What it is not is finished writing. The model has no idea what your about page should sound like, which three things you actually want to say, or which of your war stories belongs in this particular post. That work stays yours. The first pass removes the part that paralyses founders and leaves the part founders are good at: judging, sharpening, cutting, re-voicing.
Why does this matter for your business?
Because the cadence of your writing shapes the cadence of your visibility, and a founder who plans to publish weekly typically publishes quarterly. Noy and Zhang’s randomised trial in Science measured a forty percent reduction in time on professional writing tasks when AI handled the first pass and a human did the second. The gains concentrated in the drafting phase, which matches what founders report informally.
The Brynjolfsson, Li, and Raymond NBER field study from 2023 reached the same finding from a different angle. AI-assisted workers gained the largest productivity uplift when they used the model for first-pass drafting paired with human editing, rather than full automation or no use at all. For a services firm in the £1m to £10m turnover band, where the founder’s voice is the differentiating asset and the founder’s time is the binding constraint on growth, this is the rhythm that makes weekly cadence sustainable.
The compounding sits in the artefact you produce, not the tool you use. A founder who publishes a useful piece of writing every fortnight, even an imperfect one, builds a body of work that becomes a hiring asset, a sales asset, and a strategic clarity asset. A founder who sits on the fourteen-week about page builds none of those things. The first-pass-then-take-it-back rhythm is what turns a once-a-quarter intention into a once-a-fortnight habit.
Where will you actually meet it?
In four places, weekly or near-weekly, for the founders I work with. The about page or service page rewrite that has been on the to-do list for six months. The LinkedIn post that sat in drafts on Sunday night and never made it out. The client update or proposal where you rebuild the same structure from memory every time. The newsletter or blog post you committed to and quietly abandoned.
Each is a recurring writing task with a stable shape, and each is a candidate for the first-pass-then-take-it-back rhythm. The brief is what earns a useful first pass. Anthropic and OpenAI publish the same advice in slightly different language: tell the model who the audience is, the angle, the tone you want, the length, and one thing you do not want it to do. A brief like “write a thousand-word LinkedIn post for owner-managed services founders in the UK, sceptical-but-curious tone, on why the about page is the most expensive piece of paper in their week, do not list five tips” produces a draft you can work with. “Write a LinkedIn post about AI” does not.
The discipline of writing the brief is the discipline of clarifying what you actually want to say, which is the part of writing many founders skip and later regret skipping. If you find that you cannot write a clear brief for the model, that is information. It usually means you have not yet decided who the post is for, what claim you are making, or why it matters this week. The five-minute brief, written before the model touches the keyboard, is often the most valuable five minutes in the writing hour. Once that brief exists, you can keep it, refine it, and reuse it for every variation of the same recurring task.
When should you take the keyboard back, and how?
When the first pass lands, always. The cycle that works, tested across operators from Tyler Cowen to Ethan Mollick to a recruitment-firm founder I worked with last quarter, is print, mark up, dictate. Print the draft. Read it with a pen in your hand. Cut what is generic, write your own sentences in the margins, circle facts to verify. Then dictate the rewrite into your phone, reading from the paper.
The shift between media is doing the work. Mueller and Oppenheimer’s 2014 PNAS paper on longhand versus keyboard note-taking documented that handwriting engages deeper cognitive processing than typing. The same applies to editing. On screen, you skim and accept. On paper, with a pen, you notice the soft phrases, the missing emphasis, the line that sounds nothing like you. Dictation then re-introduces your spoken cadence, which is closer to your written voice than the model’s average-of-the-internet cadence will ever be. The output is yours, the model removed the friction.
Sometimes the right move is to scrap the draft entirely. Three signals: the model invented the angle you did not brief for, the tone is so far off that re-voicing every paragraph would cost more than rewriting, or the draft contains fabricated facts that force you to verify and rewrite three sections. The cost of a rejected first pass is thirty seconds. The cost of editing a wrong draft is an hour. Notice which one you are about to spend, and choose accordingly.
Related concepts
The brief is the load-bearing piece, and it deserves the same care you would give a brief to a contractor you have not worked with before. Vague briefs produce generic drafts whoever is writing. The editor’s-eye review move covers what to do once a draft lands: reading AI output the way an editor reads a junior writer, with a pen rather than a like button.
Both sit inside the wider practice of using AI on your own desk rather than buying it as a team platform. That is the rhythm a founder can adopt this week without procurement, integration, or change management. The brief earns the first pass, the print-mark-up-dictate cycle finishes it, the cadence does the compounding. If you want to book a conversation about which weekly writing tasks are the right ones to start with, that is where the work usually begins.
