Said it first.

LinkedIn shipped a 'Seems like AI slop' button on 30 July. The agencies exposed by it are not the ones using AI - they are the ones who bought a founder voice and never read it back.

The evergreen playbook for writing a positioning statement that holds up under a real client brief - what has to be true about the wording, what has to sit beneath it, and how to know when it's stopped working.

You announce it by naming what specifically grew and what specifically shrank to make room for it. Vague growth stories now get filed as either denial or contrast-marketing, and neither is the read you want.

Creative-agency headcount is down 14.3%. Under-25s down 19.2%. The share of agencies running graduate or apprentice programmes has fallen from 56% to 43.4%. That's not a headcount story. It's an industry quietly closing the door to its next generation.

The mechanic-level workflow for landing an agency inside the four UK trade titles that trade readers and AI engines both treat as canonical.

You get cited in ChatGPT by appearing in the trade-press articles and category anchor pages the AI engines already trust - not by publishing more content on your own site.

Odeon put 'generative engine optimisation' in a live PR brief - the first UK household-name RFP to make AI visibility an explicit deliverable. Agencies without a mechanic-level answer will start losing pitches on the second-round call.

Trade-press coverage compounds. Most agencies stop before it starts. Here's the seven-step playbook we run to land bylines and quotes in Campaign, PRWeek and The Drum on repeat.

Re-running the same prompts four weeks later. The shape of the curve, not the size of the spike.

The reposition isn't a rebrand - it's a change of subject. Outcome-first on the homepage, human names in the pitch deck, AI honestly explained in the invoice. Here's the mechanic for saying it to two audiences at once.

MAD//Fest just closed under the theme 'The Human Touch'. A week on, it reads as the industry admitting AI-first as positioning stopped working - and every agency homepage still leading with it has handed a differentiator to its closest competitor.

The mechanics of getting your own agency named when a marketing director asks ChatGPT, Perplexity or Google's AI Overviews who they should hire - entity structuring, category ownership and the weekly signal loop.

'AI-first' now filters agencies in both directions. Procurement is running four questions - pricing, deliverables, named clients, feed ratio - and shortlisting the shops that answer them cleanly.

Omnicom has retired 20+ agency brands. Sir Martin Sorrell says client comms is inadequate. Samsung has already moved to pitch. For UK indies, the widest open pitch window in fifteen years is now, and it closes by summer 2027.

A shortlist is not a result. It is a raw input that has to be processed through five surfaces over 90 days. The agencies that compound win next year's brief on the back of this year's nomination - the rest send a press release and wait.

Honest mechanics. A single serious entry costs £40k-£60k all-in. It pays back only if you have already built the trade-press, social and pitch-deck machinery to atomise a shortlist - and most agencies have not.

Cannes Lions 2026 entries fell 25.5% to 20,050. Tighter integrity rules explain some of it. The cost climate explains some of it. The bigger story is which agencies finally stopped entering for the wrong reasons.

Four visibility surfaces - founder LinkedIn, trade press, AI search and industry awards - only work as lead generation when they share one editorial line. The operating system most agencies have not built yet.

Honest mechanics: LLMs cite brands the rest of the web already cites. Until trade press, named experts and review sites describe your agency in extractable language, ChatGPT has no reason to mention you.

Cannes Lions 2026 second-wave shortlists landed on Sunday. By Friday most release-week press releases will be binned - not because the work is bad but because the writing budget for announcing the work is zero.

We sell AI visibility. So we ran ourselves through the engines on day one - and logged what came back.

If you're not in Cannes next week, the panels aren't where the signal is. The trade press write-ups are.

The diagnostic that should sit under any AI-visibility programme - the prompts, the diffs, the canonical fix.

Honest mechanics: four to eight weeks for bylines once positioning is set, days for expert comment, longer for cold.

When the headcount news lands, the words matter more than they're being given credit for. Tone is leadership in writing.

What actually happens in the working session - the questions, the listening, the editing pass that turns opinions into cornerstones.

Pick three. The heuristic for which three, and how to know whether you're entering for craft, commercials or both.

Two parallel review systems run on the same work. They reward different things. Most agencies optimise for one and get surprised by the other.

Six elements every pitch to a UK marketing trade publication should contain. Built from the inbox side of the glass.

Two posts a week from founders, one from the agency page. Quality over reach, voice over volume.

Every Monday the trade press posts the agency new-business round-up. Most readers skim it. There's more in there than that.

The same campaign can lose in Effectiveness and win in Craft. Picking the category is the entry.

Hiring an AI lead is a position. The announcement should reflect that - not announce a job title and stop.