The Power of And

When you fan away the hot air, you'll see that we really are still only at the beginning of the beginning.

Anthropic published a report this week measuring AI's actual impact on employment. The headline finding will surprise you.

No systematic increase in unemployment. Not yet.

The researchers studied workers in the most AI-exposed occupations (computer programmers at 74.5% task coverage, market research analysts at 64.8%, customer service representatives at 70.1%) and found no significant rise in joblessness since ChatGPT launched in late 2022. The data is clean, rigorous, and almost immediately buried under the weight of the one finding that follows it.

The hiring of younger workers into those same exposed occupations has already slowed by 14%.

The jobs aren't disappearing yet. The pipeline into those jobs is already closing.

That distinction matters. Not gone. Not yet visible in unemployment statistics. But structurally altered by a cohort effect, in the way that a river changes course slowly, methodically, until the day someone notices the old bank has been dry for months. What looks like a steady state is already something different.

I want to resist the framing that everything happening to our industry right now is an AI story. It isn't. And that framing leads to AI answers that are necessary but nowhere near sufficient.

The five forces I have been tracking as The Great Compression (budget decline, platform disintermediation, AI automation, talent crisis, margin compression) are all accelerating simultaneously. Omnicom is digesting IPG, eliminating thousands of positions, with my prediction of cumulative reductions looking increasingly conservative with every quarterly result. WPP is executing Elevate28 on a balance sheet with no margin for error, burning potential energy faster than strategy can replenish it. Dentsu is mid-restructure. In-housing has moved from trend to standard operating procedure at major advertisers. Marketing services budgets as a percentage of revenue have declined from 9.1% to 7.7% in two years, and the direction of travel hasn't changed.

AI is accelerating all of these forces. But it did not cause them and is being used by some as a convenient scapegoat for mismanagement.

The jobs being eliminated at Omnicom are not casualties of AI. They are casualties of a merger designed to solve compression with more compression. The junior planners and account coordinators who cannot find entry-level agency work are not being displaced by language models. They are being displaced by a structural model that no longer needs the middle layers that once defined career progression in this industry.

If you treat this as an AI problem, you will wait for an AI solution. There isn't one. The response must be as structural as the forces causing it.

Which brings me to the choice you actually face.

I have written before about what happens when powerful winds blow through an industry, or through a team, or through an individual career. The response, consistently, falls into one of three categories.

The first is to let the wind blow you around.

To be a victim of wherever it deposits you. To wait and see, to hope circumstances resolve in your favour, to treat the disruption as something happening to you rather than something you are in a relationship with. This is not passivity, in fact it is often accompanied by a great deal of activity, commentary, discussion, impressive presentations at industry conferences about embracing change. But the organisation chart stays the same. The business model stays the same. The talent strategy doesn't change. The leaf travels a long distance and arrives battered, directionless, and no closer to where it needs to be.

The second is to bunker down.

To make the bunker bigger, deeper, more impressive. To merge with another bunker and call the combined structure resilience. To insist that what you do is uniquely human, that relationships will protect you, that the technology will plateau before it reaches your role, that AI can grade homework but it cannot manage a classroom. There is often a grain of truth in these arguments. There is rarely enough truth to matter. The winds don't respect the size of the bunker. They just keep blowing.

The third is to build a windmill.

To look at the same force that is damaging everyone around you and ask a different question: not how do I protect myself from this, but how do I harness it? Wind doesn't just blow things down. It fills the sails that drive things forward. The organisations and individuals winning right now are not the ones who predicted disruption most accurately. They are the ones who turned it into propulsion.

At the individual level, in 2026, building a windmill means something specific. It means asking not will AI take my job? but what is AI liberating me from?

Most organisations are deploying AI under what I think of as the tyranny of OR. AI or the strategist. AI or the analyst. The implicit frame is substitution: one replaces the other, the cheaper option wins, the headcount reduction appears in the next quarterly synergies announcement.

That framing serves CFOs building synergy cases for boards. It doesn't serve organisations building durable competitive advantage.

The power of AND is different. You plus AI, doing work that neither could do alone. Your judgment amplified by AI's processing capacity. Your creative instinct accelerated by AI's executional speed. Your strategic synthesis informed by pattern recognition across data sets you could never manually read. The windmill, at the individual level, is the AND. It is the deliberate choice to use AI not as a replacement for your capability but as the mechanism that clears the scaffolding around it.

Here's what I mean by scaffolding.

When I worked agency-side I observed our most talented strategy people spent the majority of their working hours on work that surrounded their real thinking rather than constituting it. Writing the deck rather than building the argument. Formatting the research report rather than interrogating the finding. Managing workflow rather than directing outcomes. Compiling performance data rather than interpreting what it meant.

The thinking was exceptional. The ratio of thinking to everything-else was not.

AI is, for the first time, genuinely capable of handling the scaffolding. Not the judgment, not the relationships, not the original strategic insight, but the infrastructure of documentation, formatting, synthesis, and iteration that has always consumed the hours between decisions. The question is whether you use that recovery of time to go upward or sideways.

Sideways is doing 2025's work at 2x speed. It is efficiency without transformation. It is the leaf moving faster in the same wind.

Upward is doing work that 2025 couldn't have done at all. More original analysis. Deeper client engagement. Genuine intellectual leadership, not synthesis of others' thinking but actual interrogation of the questions that matter. The AND, applied deliberately.

The report distinguishes between "theoretical capability" (what AI could theoretically do) and "observed exposure" (what it is actually doing in professional settings today). The gap is enormous.

Across computer and mathematics occupations, theoretical AI exposure is 94%. Observed reality is 33%. AI is currently covering a fraction of what it will eventually cover. The gap closes over years, not months, and as it does, two types of worker will emerge on opposite sides of it.

Workers who spent those years competing with AI on its own terrain (faster processing, broader synthesis, lower-cost output) will have discovered they cannot win that race. The machine always improves faster than the human trying to match it.

Workers who spent those years operating above the terrain AI occupies (building irreplaceable judgment, deepening relationships, pursuing the strategic and creative work that requires human aspiration and context) will have built something that the observed exposure statistics cannot measure, because it is precisely what those statistics exclude.

Building the windmill at an individual level is a practice with specific starting points.

First: audit your time for scaffolding. For one week, track how your hours actually divide between work that requires your specific judgment and work that scaffolds around it. For most senior marketers, the honest answer is uncomfortable.

Second: define your irreplaceable contribution explicitly. Not "strategy" or "creativity" as those are categories, not capabilities. What specific contribution do you make that requires the accumulation of your particular experience, relationships, and judgment? The more specifically you can articulate it, the more deliberately you can protect and develop it.

Third: use AI to go upward, not just sideways. The default deployment is lateral: same work, faster. The powerful deployment is vertical: use the time AI returns to do higher-order work you never had capacity for before. More original research. Client relationships that go deeper. Intellectual leadership that earns authority rather than assuming it from title.

The question is simple, even if the answer takes work: when you strip away the scaffolding from your current role, what is underneath that AI genuinely cannot reach?

That is the thing worth building toward. Your windmill.

Brandflow is written by Justin Billingsley, who has spent his career on all three sides of the industry's table: senior client, global agency leader, technology founder. First published 9 March 2026 in the Brandflow newsletter on LinkedIn.