By Julia Arenson, Director, In-House Agency Review
There’s a pattern among the major brands I’ve worked with over the past seven years. They made structural decisions: built in-house teams, appointed agency partners, brought in specialist suppliers, and called that an operating model. It wasn’t. It was an org chart with a budget attached. The system underneath, how workflows, who decides what, what good looks like at each stage, was never designed. It was inherited, copy-pasted, or ignored.
We’ve ended up in a kitchen with all the ingredients, but no idea of the recipe and no clear head chef.
When AI arrived, we suddenly started asking: how does this work, and who owns it now?
That’s the wrong question. The right one is: why didn’t you know the answer before AI showed up?
The problem is orchestration, not technology
When a brand tells me their biggest problem is brief quality, I know something deeper is wrong. Poor briefs are a symptom. The real issue is that nobody has defined how different types of work should flow: who leads, who supports, and who owns decisions at each stage.
Without that clarity, in-house teams execute briefs they were never consulted on, rather than shaping the work they’re uniquely positioned to own. They become a service desk instead of a strategic asset. Agencies fill the void with whatever they can get their hands on. Everyone gets busy. Almost nobody gets effective.
This isn’t a hybrid model problem. It’s an operating model problem and exists regardless of how a brand has structured its internal and external teams. Fully in-house operations fall into the same trap: undefined workflows, inherited processes, accountability that exists on paper and disappears under pressure.
AI doesn’t fix a broken operating model. It scales it.
Not all work is created equal
Operating models fail when they treat all creative work as interchangeable. The same brief template, approval chain and timelines, whether someone is commissioning a brand platform or requesting a banner resize. The work is not the same. The system cannot be the same.
The framework I use defines four distinct work types, each requiring a fundamentally different approach.
Origination is new strategy, new creative platforms. This is where effectiveness really matters. It needs senior thinking, space and time. Rushing it is expensive. Making it routine is fatal.
Creative adaptation: takes an existing strategic direction and builds new assets from it. A copywriter, a designer, a clear brief. Not a six-week process. Not an agency pitch. A copywriter, a designer, a clear brief. Speed and clarity, not reinvention.
Production adaptation covers resizing, versioning, localisation. Pure efficiency. The brief should be so prescriptive a machine could execute it, and increasingly, it can. This is where automation earns its keep, because the parameters are fixed and there’s nothing to interpret.
Always-on content covers continuous output across social, digital and CRM. Calendar-driven, channel-led. It’s consistently the least defined workflow, and without a clear operating model becomes a constant reactive scramble.
Most brands run all four through the same funnel. Same template, same stakeholders, same timeline assumptions. That’s not a model, it’s a traffic jam, regardless of team size, whether your agency is retained or project-based, where you’ve invested in AI tooling or not.
Where AI actually fits
The brands getting this right are doing the unglamorous foundational work first. Mapping workflows by work type. Defining accountability, not just who does the work, but who’s informed and who owns the decisions. Building briefing processes that flex based on what’s being asked for.
Then they ask where AI fits.
That sequence matters. AI briefing tools that score completeness are only useful if you’ve agreed what a complete brief looks like for each work type. AI localisation tools only work if your production workflow is clear enough to automate (and don’t get me started on your DAM and Taxonomy, that’s a whole other piece). Layer AI onto an undefined system and you don’t get efficiency. You get faster chaos.
The partnership question nobody wants to answer
The operating model question doesn’t stop internally. It runs straight into the agency relationship, and that where most brand hit a wall they pretend isn’t there.
Brands have been quietly shifting work in-house for years, prroduction first, then content, increasingly strategy. But they’ve done it by subtraction, removing scope without redesigning how the two sides work together. Same structures, same commercial models, less actual work.
Agencies, in turn, have responded by pitching AI as a service upgrade rather than clarifying what they’re actually for.
The answer, when you’re honest, is primary thinking. Strategic framing. Creative origination. The work that requires judgment, not execution. Agencies that are serious about the next five years are getting specific about where they add irreplaceable value and letting go of the rest. The ones that aren’t will keep losing ground to in-house teams that are faster, cheaper and closer to the business.
The real conversation I’m having with marketing leaders isn’t “should we use AI?” It’s “how do we redesign who does what, human and machine, internal and external, so the whole system delivers both efficiency AND effectiveness?” That’s an operating model question.
Build the foundation first
The brands that will reap the rewards of advancing tech in the short term aren’t the ones with the most sophisticated AI stack. They’re the ones that did the hard work of designing how creative output gets made: clear roles, differentiated workflows, the right work going to the right team at the right time. That applies whether you are running a ten-person in-house team or a global network of agencies and production partners.
AI will make good operating models extraordinary. But it will make bad ones unbearable.
The question isn’t whether your brand is ready for AI. It’s whether your operating model is ready for anything at all.
