AI in agencies is moving beyond experiments with copy, images, and internal productivity. The more important shift is infrastructure. Large holding companies are trying to build systems that connect data, creative production, media execution, governance, and client accountability. That is a very different challenge from giving teams another tool. It is closer to rebuilding the operating layer of the agency model.
The latest Cannes discussions around holding-company AI leaders make the tension visible. Agencies want the speed and scale of AI, but they also need brand safety, interoperability, privacy discipline, measurement logic, and human judgment. For clients, the question is no longer whether an agency uses AI. The useful question is how the agency controls AI across the work that actually affects brand and revenue.
Tool adoption is the easy part
Most marketing teams can adopt individual AI tools quickly. A strategist can summarize research, a creative team can explore variations, a media team can speed up reporting, and a content team can produce drafts. Those gains are real, but they do not automatically create a better operating model. In fact, if every team adopts separate tools without shared rules, AI can increase fragmentation instead of reducing it.
Holding companies face this problem at a larger scale. They need common systems that help teams use client data safely, reuse approved assets, document decisions, avoid duplicated work, and connect creative outputs to media and measurement. That requires architecture, not only enthusiasm. The competitive advantage may come less from who has the flashiest demo and more from who can make AI reliable inside daily client work.
Governance becomes part of the product
The strategic implication is clear. AI will not remove the need for agencies, but it will expose weak operating models. Teams that rely on manual coordination, disconnected tools, and undocumented judgment will struggle to prove value. Teams that build shared infrastructure, clear approval logic, and better measurement loops can make AI a practical advantage rather than a novelty.
For brandformance leaders, the useful takeaway is to ask agencies for the system, not the slogan. The strongest partners will be able to show how AI connects to data permissions, creative governance, media decisions, and business reporting. That is where the next agency differentiation is likely to be built.
Source: Digiday – Media Buying Briefing: The holdco tech heads expound on the ups and downs of building AI
