Turn insights into solutions

Everyone's been arguing about whether AI will take your job. Turns out most people are just using it to fix their emails.

Anthropic's dataset is rare because it shows behaviour, not intentions. No surveys, no "how do you feel about AI at work" but 81,000 conversations. What people report to the AI bot they do, overwhelmingly, is ask for help with writing, coding, and summarising things. Not autonomous agents running entire workflows.

Every technology that eventually rewired how we work entered through the same door: low-risk, easy-to-reverse tasks. The spreadsheet didn't arrive as a strategic transformation — it arrived as a slightly better way to do what accountants were already doing. Then it quietly changed what was expected of every accountant.

The more structurally telling finding is how uneven adoption is. Knowledge workers are being reshaped first — and fast. Everyone else, largely untouched. The "AI revolution" isn't sweeping through organisations; it's pooling in specific roles and functions, accumulating quietly.

But once enough micro-uses stack up, they shift the baseline. What used to take a day takes an hour. And suddenly the expectation recalibrates, but because enough small recalibrations already happened. First quietly, then all at once.

For twenty years, brand strategy was basically a funnel with good typography. Search → click → experience → convert. You owned the journey. You designed the touchpoints. You A/B tested the button colour.

That sequence is quietly being replaced by: prompt → summary → selection. And the brand experience — the thing you spent years crafting — gets compressed into a sentence an AI generates about you. You don't write it. You don't design it. You just hope the training data was kind.

When an agent decides which product to recommend, what it's drawing on isn't brand perception — it's whatever is verifiable, structured, and consistently present across the sources it was trained on. Trust shifts from feeling to data integrity. Not what your brand says, but what systems can confirm. Which means the brief for brand content quietly changes: you're no longer writing for humans discovering you — you're writing for machines that will summarise you to humans.

The SEO-to-GEO shift Bain describes is real, but the deeper implication is more unsettling than it first appears. Brands never fully controlled how they were perceived, summarised, or recommended. Word of mouth, editorial coverage, retailer placement — these were all mediating layers brands influenced but didn't own. What's changed is not the loss of control but the loss of the illusion of control. The funnel was always a fiction brands told themselves. The agent just makes the fiction harder to maintain.

The brands best positioned for this future are not the ones most invested in being loved. They are the ones most invested in being legible — consistent, verifiable, structured enough for a machine to trust. That is a different brief from anything brand teams have been given before. And it rewards qualities — clarity, consistency, data discipline — that have never been particularly glamorous.

Pharmaceutical advertising is commonly structured around the solution. This film draws a deliberate parallel between athletic training and scientific research — disciplines of iteration, of trying again after failure, and never declaring the work complete. The voiceover, taken from a 1954 educational film, is flat and almost bureaucratic. There is no swelling music, no patient testimonial, no moment of triumph. The edit moves through footage without hierarchy or climax. The creative approach celebrates persistence over breakthrough. Trying and failing. Learning. And then trying again.

Lilly spent much of 2025 in the Oval Office negotiating drug prices under political pressure, signing "most favoured nation" deals with the Trump administration to bring down the cost of its obesity drug Zepbound, while simultaneously fighting compounded versions of the product flooding the market. It is one of the most commercially and politically exposed companies in America right now. The conventional pharma ad — miracle drug, transformed patient, small-print liability — would have invited exactly the scrutiny the company is trying to deflect. By removing the product entirely, Lilly also removed the target.

What remains is a disposition claim: we are the kind of company that understands science takes time. But notice what the chosen sonic register does to that claim. The 1954 voiceover doesn't just sound old — it sounds specifically American, specifically Eisenhower-era, specifically pre-irony: earnest, institutional, a country that trusted its experts without showboating about it. Whether consciously or not, an ad about pharmaceutical patience ends up rhyming perfectly with the aesthetic of a particular national nostalgia. America when it was serious and undistracted, even if nobody in the room intended it that way.

Wieden+Kennedy is the agency that in 2018 pushed Nike to put Kaepernick in an ad, at a moment when doing so meant burning shoes and death threats and thirty-one percent online sales growth. That campaign said: advertising should take a side. "Never Over" says: we are going to take absolutely no sides at all in the most elegant way we know how. Both are sophisticated readings of a political moment. In 2026, even the most progressive creative voice in advertising has calculated that to sound like an educational film from the Eisenhower administration is just fine. Nobody seems to have noticed. That silence is also a signal.

Emily Dickinson used em dashes compulsively. So did Jane Austen. So, apparently, does your AI assistant — and that's now everyone's problem.

The episode opens with a journalist accused of using ChatGPT. The evidence? Em dashes. Not hallucinated facts, not robotic phrasing, not suspiciously perfect grammar. Just a punctuation mark that's been a staple of expressive writing for centuries. The accusation says a lot about the moment we're in: one where style is no longer interpreted, it's forensically scanned.

AI didn't invent the em dash aesthetic. It learned it — from Dickinson, Austen, from every breathless human writer who ever needed a pause that wasn't quite a comma and wasn't quite a full stop. The machine absorbed the pattern, reproduced it at scale, and now the original owners of the style look like the copies.

This is the feedback loop nobody planned for. Human traits, once learned by machines, return as markers of artificiality. Write too naturally and you sound like a bot. Overexplain and you sound like you're trying to prove you're not a bot. There's no clean exit.

The episode's most revealing moment is the "AM dash" — a speculative, human-only punctuation mark proposed as a kind of stylistic passport. It's an accidental confession: we've started treating writing as a pattern-matching problem rather than an act of thought, intention, and effort.

The headline McKinsey wants you to take away: employees are ready, leaders are dragging their feet. Neat, actionable, perfect for a C-suite audience that needs a nudge. The actual data is less tidy.

Yes, employees are using AI regularly. Yes, they're more aware of its impact on their work than leaders assume. But 41% of the workforce remains apprehensive — not a footnote, nearly half. Trust concerns around accuracy and cybersecurity are widespread. And "using AI" in this context mostly means summarising. Helpful but hardly transformational.

The 1% maturity figure is the one worth sitting with. After years of investment, breathless coverage, and organisational pilots, one in a hundred companies feels AI is actually integrated into how they work. That's not a leadership problem. That's a signal that the gap between tool adoption and structural change is much wider — and stranger — than the substitution-or-augmentation debate captures.

What's missing from the McKinsey frame is that readiness isn't binary. Workers aren't simply ready or not — they're adapting unevenly, by role, by generation, by how much their job actually changes when AI enters the room. Millennials in managerial positions are apparently the missionaries here, bridging enthusiasm and anxiety across teams. Which suggests the real unit of change isn't the organisation or the individual, but something messier in between.

The real constraint the data reveals isn't leadership or culture or investment. It's imagination. Most people are using the most powerful cognitive tool in history to do slightly faster versions of tasks they were already doing. The gap between what AI can do and what people are asking it to do is not a technology problem. It's a problem of not yet knowing what to want.