Published March 17, 2025 · By CampaignsLive · Industry
The tooling side of the AI creative shift has been well-documented. The org-chart side has not. What actually changed inside agencies through 2024 — the roles that disappeared, the roles that emerged, the workflows that reorganized around them — is a quieter story but probably the more consequential one. The tooling will keep changing; the operational structures that wrap around the tooling are slower to shift and stickier once they have.
This is a working description of what the inside of an agency looks like differently at the start of 2025 than it did at the start of 2024, drawn from observation rather than from the marketing communications side of the same agencies.
What contracted
Three categories of work absorbed visible reductions through 2024.
Junior production design. The roles whose primary work was producing the long tail of campaign variants — format adaptations, localizations, asset retouching, comp work, paid-social variants — were the most directly affected. The category that historically absorbed first jobs and trained the next generation of art directors and designers shrank materially. The agencies that ran significant performance and paid-social work were the most affected; the agencies that ran primarily editorial and brand-equity work were less so.
Stock photography and licensing. The internal teams that managed stock libraries, ran image licensing, and handled the operational overhead of sourcing imagery from external libraries reduced significantly. The shift was structural: the work AI-generated imagery replaced was, in many cases, work that had been done by sourcing and adapting stock photography. Cutting the source-and-adapt step also cut the team that managed it.
Internal retouching. The cleanup, color correction, environmental fill, sky replacement, and detail work that historically employed mid-level retouchers either moved into AI-assisted workflows that required fewer people or moved into the AI generation step itself. The most experienced retouchers retained their positions because the work that remained — high-stakes hero retouching, talent-sensitive cleanup, brand-consistency enforcement — still required them. The mid-level tier compressed.
What expanded
Three categories grew, in some cases offsetting the contractions and in some cases not.
AI-aware art direction. Senior creative roles that combined traditional art direction with the ability to brief, evaluate, and direct AI-generated work expanded. The skill profile shifted from pure visual craft to a hybrid that included prompt construction, model selection, and the editorial judgment to use generative output well. Several agencies created explicit roles for this hybrid — variously titled AI Creative Director, Generative Lead, Director of Creative Technology — that consolidated the work that had previously been distributed across more junior teams.
Production engineering. Pipeline-engineering roles, traditionally rare in agency org charts, became common. The work involves integrating generative tools into existing production workflows, building internal interfaces on top of the commercial AI products, managing model fine-tuning workflows, and operationalizing the rights and compliance infrastructure the SAG-AFTRA and equivalent agreements had set up. The roles look more like software engineering than like traditional creative production, and the agencies that grew them most aggressively hired from outside the industry.
Strategy and brand-AI consulting. Account-side roles whose work involved advising brands on AI adoption — what to use AI for, where to draw the line, how to communicate the choice to audiences — emerged as a distinct function. The work had been distributed across strategy, account management, and creative leadership in 2023. By the end of 2024 it was beginning to be its own line on the org chart at the larger agencies.
What got reorganized rather than added or removed
Two categories shifted shape without growing or shrinking dramatically.
Project management. PM workloads did not contract — if anything, they grew — but the shape of the work changed. Project managers in 2024 spent meaningfully more time on AI workflow coordination than in 2023: tracking which assets were generated, which were augmented, which were traditionally produced; coordinating with the legal and rights side on talent scanning, model provenance, and disclosure; running the operational steps that the AI workflow had introduced. The job got harder, not easier, even though some of the underlying production it tracked got simpler.
Quality assurance. Internal QA for creative deliverables shifted from primarily-aesthetic review to a more layered review that included AI-derivation checks, hand and face fidelity, identity consistency across formats, and the compositional failure modes specific to generative work. The work was the same shape as before — looking carefully at finished assets and flagging problems — but the specific failure modes had changed enough that the role required new training.
Where the friction has been
Three places.
Talent retention at the senior creative level. The hybrid AI-aware art direction roles are competitive enough that the senior creative people who can fill them have multiple options. Agencies that have struggled to retain this group have done so for predictable reasons: rigid org structures, limited ability to invest in tooling, slow procurement that prevents the senior creatives from using the tools they want to use. The agencies that have retained well have generally been the ones that gave the senior creatives operational latitude to shape their own workflows.
Pipeline integration at scale. The production engineering roles are individually productive but slow to scale, because the work is bespoke to each agency’s existing workflow. The integration cost is high and the resulting infrastructure is non-portable between agencies. Agencies that tried to standardize pipeline engineering early have produced more brittle systems than agencies that let the function grow organically.
Cost-of-change vs. cost-of-not-changing. The agencies that moved most aggressively on the restructuring have been the ones whose senior leadership made an explicit bet on the shift. The agencies that moved most slowly have been the ones that treated AI adoption as a tactical procurement question rather than a structural reorganization question. The cost-of-change for the first group has been visible (severance, reorganization, tooling investment). The cost-of-not-changing for the second group is currently less visible but is starting to show up in client-side procurement: brands that have completed their own internal AI workflows are increasingly skeptical of agency partners who have not.
What the new shape looks like
The agency that has come through the 2024 restructuring intact looks different from the agency of 2023 in the following ways.
- Smaller production teams, with the work that remains skewing toward senior craft and AI-aware art direction.
- A pipeline engineering team that did not exist eighteen months earlier.
- A strategy function that includes brand-AI advisory as a named capability.
- Rights, compliance, and disclosure operations that are visible in the org chart rather than buried in legal.
- Project management roles that are more operationally complex than they were.
- A QA function that has absorbed new failure-mode awareness.
The aggregate effect is an agency that is leaner in production headcount, heavier in technical and senior creative roles, and operationally more complex than the agency of two years earlier. Whether this is an improvement or a regression depends on which metric you read. Production volume per FTE is higher. Per-asset brand-equity quality is, at the best of the new agencies, comparable to before. The flow of mid-career creative talent into the industry is materially smaller, which is the longer concern.
For the related discussion of how the holding companies restructured, see WPP, Publicis, Omnicom: How the Holding Companies Bought Into AI. For the in-house consolidation side at brand level, see Klarna’s All-AI Marketing Org: A Year In.