For all the promises AI has made to marketers of faster insights, fewer manual tasks, smarter decisions, one message is becoming clearer by the day: AI won’t fix what’s already broken. In fact, layering a complex AI tool on top of a broken data foundation - or engaging in ‘AI theater’ - could end up making the chaos worse.
This warning comes from Vincent Spruyt, Global Chief Product Officer at KINESSO (IPG Mediabrands), who joined The Undiscovered Metric to share how AI is really being used in large-scale marketing operations, and what teams need to have in place before they scale it. Read on for key insights from the conversation, or watch the whole thing below.
AI won’t fix your broken systems
For Vincent, AI isn’t an automatic shortcut, it’s a catalyst. It can help teams move faster, act smarter, and scale more efficiently. But if your data foundations are shaky, it’ll just accelerate the cracks. Or, as Vincent puts it, “If you couldn't automate without AI, you cannot automate with AI.”
And that’s not just a theory. Many companies, Vincent notes, are now bolting large language models onto fragile workflows that were never built to scale. The result? More confusion, not more insight.
Without structure, AI just builds on chaos.
“If your workflows are not standardized,” he explains, “then adding AI on top of it, and especially letting your users do whatever they want with their own agent, just scales the chaos.”
Without governance, your AI becomes a free-for-all
Governance has quickly become a make-or-break factor for enterprise AI adoption. With the introduction of agentic AI. It’s not just conversational interfaces either, agents that act on behalf of users and collaborate with each other, all of these variations of AI create a real risk of losing control of your data if there aren't any guardrails to ensure proper data governance.
Vincent has witnessed this firsthand. “You have a lot of efficiency gains, yes. But you also have the risk of creating shadow pipelines.” To stop people creating opaque, unregulated processes with AI, KINESSO built a walled environment called AI Console, a kind of enterprise-grade, private version of ChatGPT that gives media planners, strategists, and creative teams the freedom to create agents, but within the structure of a governed platform.
Strong governance is the foundation AI needs to scale safely.
“You need to make sure you’ve built a platform around observability, around governance, around legal indemnification,” Vincent says. Otherwise, you risk sliding into what he calls “AI theatre,” impressive demos that look slick but aren’t built out with the goals and governance of the business in mind. And while these may win headlines, they rarely make it past the pilot stage.
Getting it right lowers the barrier of entry to insights
Still, none of this means AI isn’t delivering value. In fact, it’s already reshaping how work gets done, particularly by changing who gets to ask questions and access insights. Vincent describes it as giving marketers “a 24/7 Junior Analyst in their back pocket.”
With conversational interfaces layered on top of a single source of truth, marketers can now interpret complex models and explore performance data without waiting in line for an analyst to answer queries.
But while the interface may be democratized, the guardrails can’t be ignored. Even with models that cite sources or use retrieval-based grounding, hallucinations remain a risk. That’s why Vincent warns marketers against autopilot AI automation.
Don’t mistake automation for decision-making
Where AI shows the most promise today is in helping marketers get access to insights - and quickly, so they can ultimately decide what to do. It can streamline the repetitive stuff and even handle sophisticated operational workflows. But the strategic layer? That still needs a human brain.
“AI won't replace marketers, but marketers who use AI will replace marketers who don't,” says Vincent. Rather than putting AI on autopilot, marketers need to stay in the loop, using it as a copilot to guide decisions instead of letting AI make them autonomously.
“The strategy stays human. It's still people that are leveraging those tools,” says Vincent. That applies whether you're ingesting a 40-page brief, generating creative QA rules, or modeling TAM (total addressable market) projections.
You can offload the heavy lifting, just not the thinking.
AI isn’t your competitive edge, but a data-driven culture might be
One of the biggest blockers to effective AI use in agencies isn’t the tools themselves - it’s how companies create inflexible structures around them. Vincent sees a common pitfall across agencies and enterprise marketing teams: a desire to customize every solution for every client.
“The client thinks they want customization, right? But what they're actually asking for is personalized outcomes.” says Vincent. The difference is subtle but crucial. Customization often leads to bespoke systems: one-off data warehouses, reporting logic, or automation flows. But these become brittle over time.
Standardization, on the other hand, enables consistent infrastructure and governance, which can then be used to deliver highly personalized results. “You can only do that if you have a standard way of working.” says Vincent.
That mindset shift is especially important now that most organizations are working with the same underlying tech. As Vincent puts it, “We're all working with the same tools. We're all building the exact same thing, and we're all going to elevate ourselves with AI.”
What will actually set teams apart, he argues, is the way they operationalize change. And that starts with people. “Imagine that you have a 10,000 person organization and everyone saves maybe 10 minutes a day using a Gen AI agent. Well, that is 200 person-years a year, right?”
It’s a compelling case for training, and a reminder that tools don’t create value on their own. People do.
The real question: Are your operations ready?
In the end, AI isn’t always a shortcut. Often, it’s a stress test. Whatever weaknesses already exist in your systems, it’ll find them. But with the right foundations, including standard processes, clean data, clear governance, and a well-trained team, AI becomes one of the most powerful enablers marketers have ever had.
So, the question isn’t whether AI will transform your operations. It’s whether your operations are ready for it.
