Recently, I’ve been having many conversations with chief marketing officers and marketing executives about AI in creative generation. It’s a touchy topic, partly because I come from the creative world and still consider myself a designer, writer, and creative strategist at heart.
So when I help map their AI strategy for creative, I give real thought to the people who create, the audiences who consume it, and the quality of what gets made.
Here’s what I’ve found: The real question isn’t whether to use AI. It’s how to keep human creativity at the center while still capturing everything AI can genuinely offer.
First, I think it’s important to view the modern marketing organization as an operating system, where AI and other tools drive back-end complexity while humans lead strategy and critical decisions.
That means strategy, concepting, and creative direction stay firmly human, with one caveat: AI can still feed that process with critical insights. From there, AI’s job is amplification, especially versioning: multiplying the chosen idea into the localized, personalized versions the modern marketing machine runs on.
The hard part isn’t deciding whether AI belongs. It’s knowing exactly where in the flow to let it in, and just as importantly, where to keep it out. Treat it as a holistic operating system rather than a series of tool decisions, and the answer gets a lot more deliberate.
It starts at intake
Long before anyone opens a design tool, there’s a brief, and this is often where things go off the rails: scattered inputs, inconsistent formats, key performance indicators restated three ways by three stakeholders.
It’s also where agentic tools earn their place first, synthesizing past performance data, brand guidelines, market research, and competitive scans into a structured starting brief in minutes instead of days.
Humans still drive the synthesis; someone has to decide what the brief is actually arguing for. But AI brings a consistency and rigor to that document most creative teams have never had, and keeps the system mapped back to the KPIs that matter instead of whatever got loudest in the room.
AI’s roles in ideation and production
Human-generated ideation is the part I won’t compromise on with any executive I work with, and it’s where I’ve watched the most expensive mistakes happen. AI cannot generate the idea—that’s where taste, cultural read, and risk-tolerance live. Those are all judgment calls, and judgment is still a human skill. Let a model lead concepting, and you don’t just risk a bad campaign; you risk a generic one, because the model draws from the average of everything already made.
AI can accelerate the process, mocking up ten directions in the time it takes to build one, but a person still has to know which one actually says something. More often, AI’s real power is extending human creativity in ways that weren’t previously possible.
Once a concept has human sign-off, the calculus changes. Producing assets, copy drafts, image variations, and formatting for a dozen channels is exactly the high-volume, well-defined work AI handles well, because the creative decision is already made. Nobody is asking AI to decide what the campaign says here, only to help build what’s decided, faster and cheaper.
Where AI actually multiplies the work
This is the layer most executives underestimate, and the one that matters most. A single human-approved idea now has to exist in dozens, if not hundreds, of languages, formats, and audience segments, often personalized to a customer’s stage in their relationship with the brand, exactly the amplification AI is built for.
It’s already running underneath products people use every day. The idea is human. The scale is AI, and the performance data coming back should flow straight into the next brief, closing the loop.
One caution: A framework like this can calcify into dogma if you let it. The stages here are a starting map, not a rulebook, and the moment any of them run on autopilot because “that’s the process,” the system has failed at the thing it was built to protect.
That’s the model: Not AI instead of creative judgment, and not creative judgment slowed by fear of AI, but a system where the two are deliberately separated by stage, each doing the part it’s actually good at.
The organizations getting this right aren’t the ones with the most AI in their stack. They’re the ones who know exactly which stage it’s allowed to touch, and which ones belong to the people doing the creative work in the first place.
Gabriel Bridger is the global lead of experience, marketing and AI transformation at Genpact.