AI and Agencies
For years, ad agencies have been navigating disruption, from the rise of digital, to the in-housing wave, to the growing demand for performance over perception. I experienced it for a decade, being on the forefront of the rise of digital marketing. In recent years, I’ve been an outside observer of the agency world (and occasional advisor). While a few pages are similar, I don’t see the latest shift as just another chapter in the same story. It’s a redefinition of the model entirely, driven by agentic systems—AI-enabled processes where autonomous agents take on complex, multi-step workflows.
This is not just a new toolset. It’s a new operating model. And it has implications across the entire agency ecosystem, especially in media planning, buying, and creative development.
From Service Model to Systems Model
Traditional agencies are built on linear, human-driven workflows. Strategy creates the brief. Creative produces the assets. Media plans the buys. Analytics reports after the fact.
In agentic models, these handoffs become loops. AI agents can co-develop briefs, iterate creative based on performance data in real time, and autonomously adjust media allocations across platforms based on evolving signals. Agencies move from project-based producers to adaptive systems designers.
The Media Function Gets Rewritten
Creative often gets the spotlight in AI discussions, but media is where agentic models are already quietly reshaping the fundamentals.
Predictive planning: Agents trained on past performance, seasonal trends, and market conditions can propose budget allocations and channel mixes faster and more accurately than traditional planning teams.
Autonomous optimization: Instead of waiting for weekly pacing calls or monthly reports, agentic systems adjust bids, placements, and creative rotations in flight, based on real-time KPIs and constraints.
Cross-channel orchestration: Agents can manage frequency, sequencing, and spend across platforms like Meta, Google, Amazon, and programmatic buys, closing the loop between strategy and execution.
This doesn’t eliminate media strategists. It evolves them. The value shifts from execution to oversight, from pulling levers to setting parameters.
Creative Doesn’t Disappear. It Becomes Modular and Measured
On the creative side, agentic models aren’t about replacing human ideas with machine-made ones. They are about scaling the impact of strong creative thinking.
Modular asset systems allow for AI-driven remixing and personalization at scale, based on audience signals, performance, and contextual data.
Taste-driven guardrails still matter. Human judgment defines the aesthetic, tone, and boundaries for what the system generates.
Performance loops mean creative isn’t static. It evolves. Messaging and visuals can be tested, refined, and redeployed without requiring new briefing cycles.
Agencies that succeed here won’t just generate content faster. They will design systems that get smarter with each campaign.
From Account Manager to Agent Orchestrator
In this new world, the most valuable agency roles won’t be those who "do the work" but those who define, train, and oversee the systems that do. Roles will shift.
Strategists become data interpreters and system tuners.
Media buyers become orchestrators of autonomous campaigns.
Account leads shift from project shepherds to performance architects, ensuring that systems align with client business outcomes.
What It Takes to Compete
Agencies that want to lead in this agentic era will need to:
Invest in infrastructure. You need more than a few ChatGPT prompts. Real advantage comes from bespoke models, fine-tuned workflows, and integrations that reflect your unique point of view.
Rethink team structure. Hybrid teams of creatives, data scientists, engineers, and strategists—not siloed departments.
Redefine value. Move from deliverables to outcomes. The best clients won’t pay for hours. They will pay for systems that learn and scale.
AI isn’t just a disruptor. It’s an inflection point. The most relevant ad agencies of the next decade will look less like production shops and more like intelligent systems integrators. They won’t just make ads. They will design living campaigns, built on agents that sense, decide, and adapt.
Agentic models don’t eliminate people. They elevate the work. But only for those willing to rethink what an agency actually is.
