Preparing for an Agentic Marketing Operating Model

There is a moment happening right now in marketing.

In nearly every conversation with CMOs and marketing operations leaders, the vision sounds the same. Agents that surface market opportunities in real time. Agents that recommend where to invest. Agents that generate briefs, orchestrate campaigns, and even produce content.

The promise is real. And in controlled environments, it is already working. But many organizations are stuck between two instincts. They know their data and workflows are not fully connected. They also feel pressure to start using AI now.

So the question becomes. Do we wait until we are ready, or do we start experimenting?

The answer is both. Just not in the way most people think.

The real problem isn’t AI. It is how work gets done.

In controlled environments, agentic workflows can move fast. A set of agents can take in raw signals, connect the dots, and produce a structured output in minutes. Insights turn into recommendations. Recommendations turn into briefs. Work is ready to move. But what happens next is what matters.

Where does that brief live?
How does it connect to planning?
How does it turn into real work?
How is it measured and improved?

In many organizations, this is where things stall. Not because the AI failed. Because the way work gets done has not changed.

This is something that is becoming clearer across the market. Scaling AI is just about adding new tools. It requires rethinking processes, decision points, and how work flows across teams. It exposes where processes are not connected.

Building the connected data foundation and experimentation must happen together.

A lot of teams assume they need to fix their foundation before they start using AI. In reality, that approach usually slows everything down. At the same time, running isolated AI experiments that are not connected to real workflows doesn’t create lasting change.

The teams making progress are doing something different. They are using real, end-to-end experiments to reshape how work happens. Instead of asking what they need before AI, they ask a more practical question. What breaks when we run this through our current process? That is how both the foundation and the operating model evolve at the same time.

What an end-to-end experiment reveals.

Take a simple use case.

Can a set of agents take raw data signals and produce a marketing brief that a team can actually execute?

The agents will likely succeed at generating the output. The real insight comes from trying to use it.

Can it be brought into a planning system like Workfront?
Does it align with how work is prioritized?
Are the decision and approval gates clearly defined?
Can creative teams act on it without rework?
Is it on brand?
Can performance be measured and fed back in?

Every point where the process slows down is not a failure. It is a signal that something in the process needs to change.

Maybe intake is too complex.
Maybe decision rights are unclear.
Maybe the workflow is not standardized.
Maybe brand guidance is not structured enough for AI to use.

This is the real work. Not just connecting systems, but reshaping how work flows through them.

The foundation you are actually building.

Over time, these experiments start to connect. What emerges is not just a better use of AI, but a different kind of system.

Insights feed strategy.
Strategy feeds planning.
Planning drives execution.
Execution connects to activation.
Measurement feeds back into insight.

In an Adobe ecosystem, this often shows up across AEP, Workfront, Firefly Enterprise, Creative Cloud, and AEM. But the real shift is not the tools. It is that each step in the process becomes structured, connected, and repeatable. Outputs are no longer static. They are inputs to the next step. That is what allows AI to scale.

Why brand intelligence becomes part of the process.

As teams start running these workflows end to end, another gap shows up. AI does not just need access to data. It needs access to how the brand actually operates.

This is where the Brand Intelligence Layer becomes important. It captures not just guidelines, but patterns. What good creative looks like. What has worked. What should not be doneWithout it, content can be generated quickly but still requires heavy correction.

With it, content becomes usable. And more importantly, it becomes consistent across teams, channels, and use cases. This is not just a creative improvement. It is a process improvement.

Brand context becomes something that flows through the system, not something that gets reinterpreted at every step.

Why planning becomes the anchor

As processes evolve, one layer becomes the anchor point.

Planning.

This is where decisions are made and turned into work. It is where AI outputs become real inputs into execution. Tools like Workfront become the place where:

Insights are translated into structured briefs.
Work is prioritized and assigned.
Execution is tracked and governed.

When this works, AI is not operating alongside the process. It is embedded within it. That is the difference between experimentation and scale.

What you learn along the way.

The biggest benefit of this approach is not just what AI produces.

It is what it reveals.

You see where your processes are unclear.
You see where handoffs break down.
You see where data is missing or inconsistent.
You see where brand guidance is too abstract to be usable.
You see where measurement is disconnected from decisions.

These are the things that prevent organizations from scaling anything, with or without AI.

And they only become visible when you run real work through the system.

Why a culture of curiosity matters more than tools

Even with the right systems and processes, one thing becomes clear. This only works if the organization is willing to adapt.

The teams that move fastest are not the ones with perfect plans. They are the ones that are willing to test, observe, and adjust. This is where a culture of curiosity becomes paramount. It encourages teams to run focused experiments, understand how the process behaves, and improve it quickly.

Not in theory. In practice.

If something doesn’t work, that is useful. It tells you what needs to change. Over time, this builds a different kind of capability. Not just execution. Continuous improvement.

The real transformation taking place.

At a certain point, this stops feeling like a set of experiments.

It starts to feel like a system.

One where:

Insights are continuously generated and acted on.
Decisions are structured and visible.
Work flows across teams in a consistent way.
Creative stays aligned because brand context is built in.
Performance feeds the next cycle.

Marketing starts to behave less like a series of campaigns and more like a connected, evolving system.

Where to start.

You do not need to redesign everything to begin. You also do not need to deploy AI everywhere.

Start with one use case. Run it end to end. Watch how the process behaves. Fix what slows it down. Run it again.

That is how both the process and the foundation improve. Not separately. Together.

Final thought

Most teams are looking at agentic AI and asking what it can do. A better question is what needs to change for it to actually work. Because the advantage isn’t just in generating outputs faster. It is in building processes that can absorb those outputs, turn them into action, and improve over time.

The teams that figure that out will not just adopt AI. They will operate differently. And that is what will set them apart.

Authored by:
Adam Driggs, Account Director, Omnicom Adobe Practice