When you survey 500 marketers, you really learn about what’s actually going on day-to-day and campaign to campaign, and here are my top 3 takeaways.
The focus of the survey was the state of AI in creative and content operations. The findings are clear: AI is playing a role in everybody’s day-to-day work. But maturity is lagging; what does that mean for me as a Marketing leader and practitioner?
Marketing has picked AI up fast, and they’re using it everywhere. The insights emerge when you question a bit deeper.
The Question: Why is everyone using AI, but it isn’t embedded?
We discovered that only 24% of organisations say they’ve embedded AI into everyday workflows. 43% describe their own AI usage as “moderate” rather than extensive, which tells us most teams already sense the disconnect in how they are operating. The question is why?
And from conversations, there are several key reasons for this:
- Leadership – our organization doesn’t allow it
- System – our way of working doesn’t support embedding agents
- Ownership – I can change the way I work, but the process is another beast
- Knowledge – I don’t know how to move forward
- Time – I want to prove it works before I embed it
However, regardless of the reason, the result is a suboptimal process. And once duplication, inconsistency, and rework creep into projects, poor ways of working with AI will become increasingly expensive to sustain.
We looked a bit deeper into the stats again to learn that the picture wasn’t concentrated anywhere in particular: reporting (27%), content generation (27%), tagging (27%), compliance (26%), and workflow automation (26%) all sit within a single point of each other. That’s an unusual kind of adoption curve, because normally, once something reaches “everywhere,” you’d expect higher rates of utilisation across the majority of use cases. It seems in this case, breadth arrived well ahead of depth, and I don’t think that’s an accident.
The learning: Using AI and embedding AI are different things
This moderate rate of AI ultilisation spread across use cases is a clue to an underlying issue.
Here’s what I think is happening for many marketing teams with AI. They pick up an AI tool because it solves an immediate problem: a faster first draft, a quicker summary, a campaign mockup in minutes. Those specific tools then sit external to the existing process, rather than inside it. The brief is being written the old way. The approval still gets routed the old way. Governance happens in the old way, if at all.
I don’t think we can pinpoint this as a maturity failure on any one person’s part. It’s what happens when adoption outpaces the operating model that’s powering the team. The whole team said yes to the tool, but almost nobody went back and redesigned the workflow around it. It’s until teams come together to implement the workflow updates that “using AI” and “AI being embedded” are going to keep meaning two very different things.
What great looks like: How does embedded AI work in marketing and creative operations?
So, what is the other version, and how does it work? When it comes to briefing a project or campaign, embedded AI doesn’t sit next to the brief. It’s part of producing the brief, with access to historical project data and previous briefs; with a click of a button, the agent retrieves that knowledge to populate the current brief. In this case, AI is streamlining the brief creation process.
The same is true for every stage; it doesn’t sit next to the approval. It’s part of how that approval gets routed, tracked, and signed off. The difference is whether the output has somewhere to go, be it a workflow that knows the next step, who needs to see it, and what “done” and “approved” mean once AI is involved.
That’s the model and capabilities we’ve built into Screendragon’s AI platform. It’s AI that runs inside the workflow a team already has, governed by the same rules as everything else, rather than a separate tool people have to remember to use properly, or manage the risk themselves. Decide on the use case, build it directly into the system, and avoid the extra work of trying to retrofit an existing AI solution into the workflow.
How to get to great: A smart, end-to-end AI way of working, embedded where work happens.
If you recognise your own organisation somewhere in that 43% “moderate” figure, progress isn’t about adopting more tools or delivering more training sessions. I believe the next step is smaller and more structural:
- Pick one business process – for example, briefing, content creation, or approval – and map how the work happens, including where AI is already being used informally.
- Define the rules around AI within that process: where it can be used, what still requires human judgement or approval, and who is accountable.
- Then embed that process into the technology people actually work in, so AI, approvals, decisions, and hand-offs happen as part of the workflow.
- Connect and trace every output back to the original brief, a specific decision and/or approval, so all work is visible, and AI can continuously learn from previous actions, decisions and outcomes.
- Once that end-to-end way of working is proven, repeat it for the next process rather than trying to roll AI out everywhere at once.
I’ve seen that a small and structural approach beats a broad and shallow approach every time. It’s a less exciting story to tell internally than “we rolled out AI across the whole department,” but it’s the version that compounds in value over time.
My final thoughts
Here’s where I land on this: marketing teams are working with AI in two ways: either AI sits alongside where work gets done, scattered across multiple tools, or it’s being used within one system, enabling the orchestration of work. Of course, there is the transition period, where it’s a bit of both.
Personally, I’d rather my own team were slower to add a tenth AI tool and further along on making sure the first five are properly embedded. That’s the trade-off I think marketing leaders are facing. Embedded AI is the future of how work gets done, so start getting your teams involved in implementing it.
If your workflow system can’t support the creation of custom AI agents, or embed the agents you already use, it may be time to question whether it is a system that will sustain your team into the future.
I’d recommend reading the full findings in our State of Creative and Content Operations 2026 report to get a full picture of where marketing teams are at in their AI adoption journey.
I’d be genuinely curious where your own organisation sits against that 24% figure. Feel free to reach out to me on LinkedIn.
CMO, Screendragon