The Great In-Housing: When Brands Discovered They Don't Need Us
As brands bring creative production in-house with AI tools that work faster than their own agencies, a generation of junior talent wonders if their career path just evaporated overnight.

The numbers landed like a gut punch at last week's agency strategy session. 32% of brands expect to handle nearly all creative in-house within 12 months, while another 23% expect to bring at least half of their creative work in-house. Sophie, a 26-year-old copywriter at a mid-tier creative shop, watched the deck unfold with the queasy recognition of someone seeing her own funeral arrangements.
"So what happens to us?" she asked during the Q&A. The room went silent.
The Production Tsunami
The in-housing surge isn't theoretical anymore. Kimberly-Clark said an internally developed AI platform reduced content creation timelines from 24 days to two hours, while companies like Catalyst Brands and Target India are using AI across parts of the marketing process including creating product images and videos, choosing influencers, and localising campaigns.
The scale is breathtaking. The intimacy, terrifying.
When M&S, one of the United Kingdom's largest retailers, needed same-day turnaround on their product descriptions, Jellyfish delivered a GenAI solution. The result: 80% faster content delivery, a 30% reduction in unit costs and a 25% boost in production capacity.
These aren't pilot programs anymore. They're replacing entire categories of work that used to require briefings, presentations, three rounds of revisions, and invoices measured in the tens of thousands.
When Strategy Becomes Commodity
But here's where the story gets more complex than the usual "robots are taking our jobs" narrative. Industry analysts expect agencies to remain important for strategy, creative direction, and specialised expertise, but production-heavy activities are increasingly being automated within company-owned operations.
Jon Cook, who now leads WPP's newly formed Creative division, calls it "not an agency. It's an operating system that lets those great agencies, those very creative agencies, operate together". That's the kind of corporate-speak that would make David Ogilvy roll over in his grave.
Brian Wieser, CEO of advisory firm Madison and Wall, said the question is whether clients still see value in expertise. He said companies can produce average work internally if that is enough, but higher-value work still depends on specialist judgement.
Average work. Let that sink in. When did "good enough" become the industry benchmark?
The Agency Response: Panic or Evolution?
More than half (54%) of agency leaders surveyed said they're somewhat or very likely to start their own firm within the next two years. The rats aren't just leaving the ship: they're building their own boats.
Building enterprise-grade AI capabilities requires significant investment. Omnicom, WPP, Publicis, and others have committed hundreds of millions to AI development. The holding companies are spending fortunes trying to build what their clients can now buy off the shelf.
WPP integrated Imagen 3 into its "WPP Open" marketing platform, enabling global clients like Verizon, L'Oreal, and Unilever to scale content production. Similarly, Shopify introduced the technology to its merchants, helping them expand product photography and generate high-quality lifestyle imagery.
So WPP spent millions building an "operating system" to help clients use Google's tools more efficiently. That's not strategy. That's systems integration.
The Junior Talent Exodus
The human cost hits different when you're 24 and just figured out how to write a brief that doesn't sound like a college essay. Marketing teams must hire prompt engineers, data analysts, and ad operations leads. Those aren't creative roles. They're technical ones.
The career ladder that once led from junior copywriter to creative director now leads where, exactly? To prompt optimisation? To AI quality assurance? To explaining to clients why their AI-generated campaign feels "off-brand"?
"AI can help in the ideation phase, like storyboarding multiple concepts quickly," says Te'Shawn Dwyer, a manager of StackAdapt's in-house Creative Studio team. "But when brands skip human curation, the results can feel cold and impersonal".
Cold and impersonal. But fast and cheap.
The Quality Question Nobody Wants to Answer
IAB reported that 70% of marketers had experienced at least one AI-related incident. These included hallucinated outputs, biased or inappropriate content, and off-brand material.
Yet brands keep building these capabilities because the economic pressure is overwhelming. Marketing budgets have remained flat as a share of company revenue. Gartner's 2025 CMO Spend Survey found that marketing budgets remained at 7.7% of overall company revenue, unchanged from the previous year.
So if you can't spend more, you have to spend smarter. And apparently "smarter" means accepting the occasional brand disaster in exchange for 80% cost savings.
Consumer surveys reveal scepticism toward undisclosed AI content. Approximately 36% of adults feel less inclined to purchase after seeing synthetic work. But CMOs aren't measured on consumer trust in their creative process. They're measured on efficiency metrics and cost per acquisition.
The brutal truth is that the creative industry built itself on labour arbitrage and is now being arbitraged out of existence. We convinced brands they needed us to access talent, manage complexity, and deliver scale. AI doesn't need us for any of that.
So here's the question that haunts every late-night agency strategy session: If brands can produce "good enough" creative work in-house, faster and cheaper than we can, what exactly are we selling them?
And what are we telling Sophie?