👷🏽 From maker to architect

Looking back at the first half of this year, I’ve done more work in process design and AI workflow implementation than in pure content strategy. Building full text-driven content workflows. Using AI as a video series co-producer: from researching and choosing guests to repurposing the finished piece into new formats.

Alongside that, I’ve seen a shift in how content work is structured.

The old model moves a piece between distinct collaborators as it progresses: the strategist directs, the writer writes, the editor reviews. The model I’m seeing more and more now is a single person owning the initiative from brief to publication — what Elena Verna recently called the High-Impact IC. A contributor who does the actual work, folds in feedback from teammates and subject matter experts, and makes sure what was planned at the start actually ships.

Both trends are pointing in the same direction. The content marketer’s role is expanding deeper into orchestration, on top of execution.

Which means the skills that built your career so far need some company.

Fair warning: I’m not going to talk about “taste.” Or “judgment.” Both have had plenty of airtime lately (fair enough), so I fear they’re gradually becoming flared at the edges and losing their bright colors from overuse. The more we say, the less they mean.

Here are three specific sets of skills I think about now.

System building: The process becomes the point

Rather than asking what to make with AI next, let’s ask how it gets made. What should this system ultimately be capable of? What does “done” look like at each stage, and who’s setting that standard?

Once you have that vision, you can work backward: defining the steps, placing the quality gates, building the improvement loop that makes the system better each time it runs.

A human belongs in this process for a concrete reason: to recognize and define a level of quality. We think through written content and feel through video content as members of an audience. We’re the ones who experience whether something connects, which is exactly why we need to be in the process of making it, not only at the end approving it.

That recognition comes from being the kind of person who consumes content carefully. A rubric can capture some of it; lived experience captures more.

Ryan Law, who directs content at Ahrefs, recently mapped out what this looks like in practice. The infrastructure for an AI-native content operation (automated audits, topical analysis, daily content updates) can now be set up agentically, by one person, in a fraction of the time it once required. What you bring is the vision of what to build and the experience to know when it’s working.

Strategic focus: From fifty ideas to a shortlist of three (or zero)

When production stops being the main bottleneck, the harder calls are about direction.

“What should we be working on?” isn’t a single question. It branches into whether to build new or improve what exists, whether to go narrow on a specific audience or cast a wider net, whether to launch something or invest in building on top of what’s already there.

One of the most underrated versions of this skill is knowing when to stop building altogether. Sometimes the most useful call is to pause on new production and spend that time making what already exists better.

AI will generate fifty ideas for what you could do next. Which three to pursue — or whether to pursue any at all — is still yours to decide.

Cross-domain synthesis: Human curiosity makes you hard to replace

AI synthesizes well. It pulls from a large body of existing work, makes connections between what’s already there, and produces output that covers familiar ground thoroughly.

The limitation is what it’s drawing from and what it looks for. It focuses on what already exists in its knowledge corpus. It converges on common answers, familiar framings, the solutions everyone else is already considering.

Your advantage is “the unexpected import” from another area.

When you read outside your industry, watch how other fields solve similar problems, and let yourself notice things that seem unrelated to your work, you bring in perspective that isn’t in anyone’s “training data.” How a museum thinks about visitor flow. What a game designer does to build narrative tension. What a D2C brand tried in a downturn. These become lenses.

As you expand your inputs, the connections will follow.

The mechanism is genuinely unclear to me. When I take time to read outside my field or sit with an interactive performance, I come back with different angles than when I stay inside marketing. The connection appears later, sideways, without a clear path. My best theory is that it’s emergent. You can’t force the specific connection, but you can create the conditions for it.

One last thing

Owning the full chain doesn’t mean working alone.

I don’t assume marketers will toil all day in a dark basement churning out content pieces like they’re sprockets and widgets.

You’re still bringing in a colleague to pressure-test a strategic call, a subject-matter expert to catch what you missed, someone who represents the audience you’re trying to reach. The collaboration shifts from hand-offs to feedback loops: one person holds the thread, which means outside input actually makes it into the work rather than getting lost between people.

***

Strong production skills are still the foundation you stand on and now they’re also what qualifies you to set the quality bar, to know what the system should look like, to find out the novel and the curious, even when you’re not the one who “handmade” everything.

Next time you finish a piece of content, ask not just whether it’s good, but what you’d build differently in the process that produced it. That shift in question is where the new job starts.

Architectural skills matter,

👌 Handpicked stories for you

Here are some interesting things I stumbled upon this week while reading, writing, researching, and being a generally curious marketing person on the internet:

💼 Three for work

How NerdWallet Built a High-EEAT Review Program That Buyers and LLMs Trust | NerdWallet built what this piece calls a “Trust Stack”: named experts on every page, published review criteria, editorial independence explicitly stated. This is what EEAT as a compounding asset looks like. The discipline is in doing all of it consistently, at scale, for over a decade.

Why Calling Yourself “The Best” Could Be Hurting Your AI Overview Visibility | Being cited as a source in AI Overviews and being recommended as a brand are two very different things. Lily Ray’s research shows that when a self-promotional “best of” listicle gets cited, the brand behind it gets excluded from the actual recommendation 69% of the time. The brands that get recommended have genuine third-party authority.

A Five-Domain Content Strategy Audit for the AI Era | This piece notes the difference between “observer content” (summaries and analysis) and “participant content” (contrarian takes and unique viewpoints). Your content program may be over-indexed on the first. A five-domain audit framework helps you identify where the imbalance sits, and where AI is making it worse rather than better.

👀 Two for fun

This historic take on cat food and more | This is a fascinating story about pet food vendors long before a neatly-packaged Purina pouch. It has Victorian flair, fascinating images, and an unexpected Jack the Ripper connection.

This brilliant photo gallery | Few things excite me more than food and photography, and the World Food Photography Awards has it all. There are also categories for wine!

🐈 Weekly cat

Are you vacation ready?