Marcus Bell runs AI strategy at Cargolink, a mid-market logistics company moving freight across North America, and for three years he did excellent work that almost no one outside his company knew about. Then a competitor's head of data science published a widely shared essay on exactly the problem Cargolink had already solved, and overnight that person became the recognized voice in the space. Marcus had done the harder work. The competitor had done the writing. This chapter is about closing that gap: why publishing is the highest-leverage activity available to a senior AI leader, and how to build a writing practice that survives a demanding job.

Why Writing Is Leverage, Not Vanity

Marcus's team built a demand-forecasting system that meaningfully cut empty-truck miles, and he could explain the approach cold in any meeting. What he could not do was point to anything a stranger could read. When the competitor's essay circulated, that person started receiving conference invitations, attention from strong candidates, and quotes in the trade press, none of which required them to have solved the problem better. In a field where reputation compounds, the difference between the two of them was not competence. It was that one had made his judgment findable and the other had not.

This is the uncomfortable truth about publishing for AI leaders: the work does not speak for itself. Writing is how expertise travels beyond the room you are in. It is the highest-leverage activity available to a senior leader because a single well-made piece can recruit engineers you never met, shorten sales cycles because prospects already trust your judgment, attract partners, and give you standing in policy conversations that shape your operating environment. Marcus was not a worse practitioner than his competitor. He had simply left his most valuable asset, his hard-won judgment, locked inside his own head where it could not work for him.

Publishing is also a forcing function for thinking, and this is the half that leaders underrate. You do not fully understand a strategy until you have to defend it in prose to a sceptical reader who cannot be persuaded by your title or your tone of voice. Many leaders discover the holes in their own reasoning only when they try to write it down, and the discovery is worth the exercise even for pieces that never ship. So the case for writing is double: it builds your reputation externally and sharpens your judgment internally. The leaders who treat it as optional cede the narrative of their own field to whoever is willing to sit down and write.

What Publishing Actually Requires

The first thing Marcus had to unlearn was the belief that publishing meant academic papers or a book deal. Those are two formats among many, and for most practitioners they are the wrong place to start because they demand the most effort before producing any signal about whether anyone cares. Publishing is any durable, public artifact that carries your thinking to an audience: a technical essay, an engineering blog post, a professional-network analysis, a conference talk written up as an article, a newsletter, a contributed opinion piece in a trade outlet, or a rigorous internal report you later externalize.

What every effective piece shares is not polish but a specific structure: a clear claim that a reasonable person could disagree with, evidence from real experience rather than generality, and a takeaway the reader can use. The claim is the load-bearing element. If nobody could argue with your thesis, you have written a summary, and summaries of what everyone already knows travel nowhere no matter how well they are written. The pieces that travel are the ones that say something the writer earned the right to say.

Marcus had a warehouse of such things and had been misfiling them. What actually went wrong in their first forecasting deployment, why they abandoned an approach everyone else was hyping, what a real production incident taught them about model monitoring: these are the raw material of writing that gets read, and he had been treating them as war stories for the pub rather than as assets. The reframe is simple and it is the practical unlock of this chapter. Every expensive lesson your organization has already paid for is a piece somebody else in the field would pay attention to, because they are about to pay for it too.

From Private Expertise to Public Voice

Publishing does not stand alone. It is the amplifier for everything else a senior AI leader does. The research your team conducts is worth more when it is written up where others can build on it. The speaking you do reaches a few hundred people in a room and thousands more when the talk becomes an article. The credibility you build through writing is what earns you a seat in policy and standards conversations, because policymakers and standards bodies listen to people who have publicly demonstrated they know the terrain rather than to those who merely assert it.

There is a sequencing logic worth noticing. Writing is usually the first public act because it is asynchronous, editable, and low-risk compared to a live stage: a weak paragraph can be revised, a weak answer on a panel cannot. A leader who has published thoughtful pieces also has a body of work to point to when a conference organizer or journalist is deciding whom to invite, which is how invitations actually get made. Marcus found that his first three essays were what led to his first speaking invitation, not the other way around. Private expertise becomes a public voice through a deliberate progression, and writing is almost always the entry point.

A Content Engine, Not Random Posts

Marcus's mistake, once he started, was treating each piece as a one-off act of inspiration. That approach produces two essays and then silence when work gets busy, which is worse than never starting because it publicly signals that you gave up. What works instead is a lightweight content engine: a repeatable pipeline that turns the byproducts of your actual work into publishable pieces, so that having something to say is a consequence of doing your job rather than a separate creative burden. Here is the model he adopted.

The unlock in stage 5 is that one substantial piece can feed a month of smaller content. An essay becomes the outline of a conference talk, the talk becomes a written recap, the recap yields three shorter posts pulling out individual points, and each derivative reaches people who would never have read the original. Publishing well is less about producing more raw ideas than about extracting full value from each good one. Leaders who burn out on writing are almost always the ones generating a new idea for every artifact instead of mining the ideas they already have.

Building the Habit Without Burning Out

Consistency beats brilliance in publishing, because reputation is built by showing up repeatedly with useful thinking, not by one viral hit that nobody remembers by the following quarter. The practical question is therefore not how to write well on a good day but how to keep writing at all through a bad month. Here is how Marcus made it sustainable alongside a demanding job.

• Commit to a cadence you can defend on a bad month. He chose one substantial piece every six weeks, not one a week. An ambitious cadence you abandon after a month does more reputational harm than a modest one you keep for a year, because the abandonment is visible and the modesty is not.
• Keep an always-open idea log. The hardest part of writing is not the writing; it is facing a blank page with no idea. He kept a running list so that when a writing block arrived he chose from twenty candidates rather than starting cold, which converts the scariest step into a selection problem.
• Draft the argument before the prose. He wrote the single claim and the three supporting points first, in bullet form, and only then expanded. This prevents the most common failure, a beautifully written piece that never actually says anything, and it makes a weak thesis obvious while it is still cheap to abandon.
• Use one honest reader, not a committee. One trusted peer who will say "this section is boring" is worth more than five who say "looks great." Committee review sands off the point of view that made the piece worth reading, and the more senior the reviewers, the more reliably it does so.
• Separate drafting from editing by at least a day. Writing and criticizing use different mindsets. Trying to do both at once is why so many pieces die half-finished, with the writer rewriting the first paragraph while the argument goes unbuilt.
• Use AI to pressure-test and outline, not to generate the point of view. A model can help him structure an argument, stress-test a claim, or find a weak transition. The moment the writing stops reflecting his actual hard-won judgment it becomes the generic content readers already ignore. The value he offers is precisely what a model cannot supply: what he specifically learned running real systems at Cargolink.

Measuring Whether Your Writing Is Working

The wrong metric for thought leadership is raw view count, because a thousand shallow views from the wrong audience are worth less than fifty from the exact people you want to reach. View counts are seductive because they are easy to collect and they move; the outcomes that actually matter are slower and require you to notice them. Marcus tracked outcomes rather than vanity numbers, using a simple quarterly scorecard.

A worked example makes the difference concrete. In Marcus's second quarter of publishing on his six-week cadence, he shipped two essays. One, on why Cargolink abandoned a popular routing-optimization approach, drew a modest 1,200 views but generated four substantive inbound conversations, including one from a senior engineer who joined the team three months later and cited the essay in her first interview. The other, a broader take on AI in logistics, pulled 6,000 views and produced almost no meaningful contact because it was too general to signal real expertise to anyone in particular.

The lesson was immediate and countable: the narrower, more opinionated piece with one-fifth the views did five times the useful work. Reach and relevance are not the same variable, and optimizing for the first reliably destroys the second, because the way to make a piece broadly acceptable is to remove everything specific enough to be arguable. Marcus stopped chasing reach and leaned into specificity, which also happened to make the writing easier, since a narrow claim from real experience takes less research than a survey of a whole domain.

Making It Real in Your Context

A year after his competitor's essay stung him into action, Marcus had published eight substantial pieces. He had spoken at two industry events, both invitations that traced directly back to specific essays. Two of his best recent hires had first encountered Cargolink through his writing. And in a policy roundtable on autonomous freight, he was in the room, because the organizers had read his work and considered him a credible voice. None of this required him to become a full-time writer. It required a sustainable engine and the willingness to put his judgment into public words.

To start your own progression, work through four questions this week rather than someday. First, what are three things you know from real experience that most people in your field get wrong or do not know? Those are your first three pieces. Second, what cadence could you honestly sustain through a busy quarter, given that a modest cadence you keep beats an ambitious one you abandon? Third, who is the one honest reader you would trust to tell you when a draft is weak? Fourth, what recent post-mortem, hard decision, or repeated question from your team could become a first piece by the end of the month?

Anti-Patterns to Avoid

Most failed publishing practices fail in predictable ways, and almost all of them are the writer protecting themselves from the risk that makes a piece worth reading.

• The unarguable piece. Writing a summary of what everyone already knows in slightly new words. If no reasonable person could disagree with your claim, there is nothing for a reader to think about and nothing for them to share.
• Chasing reach over relevance. Broadening a piece until it offends nobody and interests nobody. Marcus's 6,000-view general essay produced almost no meaningful contact while his 1,200-view specific one produced four conversations and a hire.
• The heroic cadence. Announcing a weekly newsletter alongside a demanding executive job, then going silent after a month. The visible abandonment costs more credibility than a slower cadence you actually keep.
• Committee review. Circulating a draft to five stakeholders and incorporating every comment. What comes back is consensus prose with the point of view sanded off, which is the one ingredient that made the piece worth publishing.
• Outsourcing the judgment. Using a model to generate the point of view rather than to structure or stress-test it. The moment the piece stops reflecting what you specifically learned, it becomes the generic content readers already scroll past.

Practice Prompts

These exercises build the engine rather than a single piece. Work through them in order and you will finish the week with a pipeline instead of an intention.

• Mine your war stories. List every post-mortem, abandoned approach and production incident you can still remember clearly. Beside each, write the one sentence a stranger in your field would find useful. That list is your idea log.
• Write three claims. Turn your three strongest entries into single-sentence claims a reasonable person could argue with. If you cannot construct the counter-argument, the claim is not yet a claim.
• Draft argument-first. Take one claim and write only the thesis and three supporting points in bullets. Leave it a day, then judge whether the argument stands before writing a single polished sentence.
• Recruit your honest reader. Name one peer who will tell you a section is boring, ask them explicitly for that, and make it clear that "looks great" is not the service you need.
• Plan the derivatives. For your next substantial piece, write down in advance the talk, the recap and the three shorter posts it will become, so repurposing is scheduled rather than remembered.
• Build the outcome scorecard. Set up the five signals from this chapter and record a baseline now, so that in a quarter you are measuring change rather than guessing.

Reflection

These questions are about the gap between what you know and what is findable about you. Answer them honestly enough to be uncomfortable.

• If a strong candidate, a prospective partner or a policymaker searched for my thinking today, what would they find, and would it represent my actual judgment?
• What have I learned recently that the field has not learned yet, and what has stopped me from writing it down?
• Am I avoiding publishing because I am busy, or because a public claim can be argued with in a way that private expertise cannot?
• What cadence would I still be keeping a year from now, and is it the one I would announce or the one I would actually sustain?

Glossary

• Thought leadership. Public standing earned by consistently sharing judgment that others find useful, as distinct from visibility earned by promotion or job title.
• Content engine. A repeatable five-stage pipeline, capture, shape, draft, sharpen, publish and repurpose, that converts the byproducts of your working life into publishable pieces.
• Defensible claim. A thesis specific enough that a reasonable, informed person could disagree with it, and therefore specific enough to be worth a reader's attention.
• Idea log. A continuously maintained list of candidate topics drawn from post-mortems, hard decisions and recurring questions, kept so that drafting never begins from a blank page.
• Repurposing. Deriving smaller artifacts, talks, recaps, short posts, from one substantial piece so that a single good idea reaches multiple audiences.
• Honest reader. The single trusted peer who will identify where a draft is weak or boring, used deliberately in place of committee review.

Related Lessons

Publishing is the entry point to a wider set of public-voice capabilities, each of which it feeds and is fed by.

• Research & Knowledge Creation covers generating the rigorous findings that give a piece something substantial to say.
• Speaking & Public Engagement takes the progression to its next stage, where published work earns you the invitation.
• Building Influence & Platform addresses how a body of published work compounds into standing across multiple channels.
• Building Thought Leadership Through Research connects the research practice to the publishing practice as a single system.
• Writing About AI Results goes deeper on presenting technical outcomes clearly and honestly to non-specialist readers.
• Public & Community Communication extends the same judgment to audiences beyond the professional field.

Closing

The leaders who shape how their field is understood are rarely the ones with the most impressive private work. They are the ones who were willing to think in public, consistently, over time, and who accepted that a claim someone can argue with is the price of a claim worth reading. Marcus's competitor did not out-work him. He out-published him, and in a reputation-driven field that was enough to claim the ground.

Your expertise is only as valuable as your ability to move it into the world where it can recruit, persuade and influence on your behalf. The mechanism is not talent and not a viral hit; it is an engine that turns the work you are already doing into artifacts other people can read, at a cadence you can hold through a bad quarter.

Key Takeaways

• The work does not speak for itself. Writing is how expertise travels beyond the room, recruiting and persuading on your behalf while you do other things.
• Publishing is also a forcing function: you do not fully understand a strategy until you have defended it in prose to a sceptical reader.
• Effective pieces share a structure, not a format: a claim a reasonable person could dispute, evidence from real experience, and a usable takeaway.
• Run a content engine rather than waiting for inspiration, and extract derivatives from each substantial piece instead of generating a new idea every time.
• Choose a cadence you can defend on a bad month; an abandoned ambitious schedule costs more credibility than a modest one you keep.
• Measure outcomes, not views. The narrow, opinionated piece with a fraction of the reach routinely does more useful work than the general one.
• Use AI to structure and pressure-test the argument, never to supply the judgment, because your hard-won experience is the one thing a model cannot provide.

Frequently Asked Questions

I am not an academic. Does publishing mean papers and books? No, and starting there is usually a mistake. Publishing is any durable public artifact that carries your thinking: a technical essay, an engineering blog post, a professional-network analysis, a written-up talk, a newsletter, a contributed opinion piece, or an internal report you externalize. The long formats demand the most effort before producing any signal about whether the argument lands, which is the opposite of what you want early.

What if I have nothing original to say? You almost certainly do and have misfiled it as war stories. What went wrong in a deployment, why you abandoned an approach the market was hyping, what a real production incident taught you about monitoring: these are exactly what other practitioners need and cannot get elsewhere, because they are the lessons your organization has already paid for.

Should I use AI to write my pieces? Use it to outline, structure and stress-test an argument, and to find a weak transition. Do not use it to generate the point of view. The instant the writing stops reflecting your actual judgment it becomes the generic content readers already ignore, and the specific thing you learned running real systems is precisely what a model cannot supply.

How do I know whether it is working? Track outcomes on a quarterly scorecard: meaningful inbound conversations, candidates citing your writing, invitations traceable to a piece, substantive rather than applauding comments, and whether you actually shipped on cadence. Judge by those and not by views, because reach and relevance move independently and optimizing for the first tends to destroy the second.

What if my employer is nervous about me publishing? Start with material that carries no competitive sensitivity, such as methods you abandoned, general lessons from incidents with details removed, and questions your team fields repeatedly. Writing also serves the company directly through recruiting, shorter sales cycles and partner interest, which is the argument to make. Marcus's writing brought Cargolink two of its best recent hires and a seat in a policy roundtable on autonomous freight.