AI should be a tool, not your talent
Image Credit: @gustavo
There's a version of this conversation that always turns into a polarising disagreement - AI versus agencies, machines versus creatives, efficiency versus thought leadership.
The real question was never whether AI belongs in marketing. It's already here, whether we like it or not. The actual question is where it belongs, and what happens to creative work if not used prudently.
Here's the uncomfortable middle ground: refusing to integrate AI into your marketing process isn't really a principled stand – refusing to use the tech will eventually mean falling behind on efficiency. On the flip side, blindly integrating it everywhere without sound judgment isn’t innovation either– it's merely outsourcing the strategy thinking to a system that was never built to think for you in the first place.
AI is infrastructure now, not a novelty
Within the last few years, AI quietly stopped being the shiny new tool marketers demoed at conferences and became the thing quietly running in the background of processes we see in the modern world. Media buying algorithms, audience segmentation, copy generation, first-draft concepts, sales, asset creation and even campaign reporting that used to take an analyst a full day - all of it has AI sitting somewhere in the pipeline now, whether we advertise that fact or not.
The numbers back this up: 95% of B2B marketers report using AI-powered tools in some capacity according to Content Marketing Institute's 2025 research, and generative AI use in at least one recurring marketing workflow has become close to standard practice.
Treating that as optional at this point causes us to fall behind. One technically can run a full-functioning marketing department without touching any of it - but also be moving at a fraction of the speed of everyone else in their category, spending more to get there, and still arriving late.
It’s all about balance when it comes to new tech integration, the one-step-ahead marketers among us are the ones who figured out early which parts of the job AI should own, and which parts it should never be allowed around.
Where AI earns its seat at the table
There's a long list of marketing and advertising work that was always more repetitive and mechanical than creative, and AI has made that distinction impossible to ignore.
Research and synthesis is the obvious one. Digesting a hundred pages of data, competitor insights, or consumer sentiment used to consume days out of a strategist's week before the actual thinking could even start. AI collapses that into hours, and frees up the strategist to spend their time on the part that actually requires a human brain: deciding what the data means, if it's relevant and what to do with the information when it comes to creative strategy.
Production is another issue entirely. Generating initial copy structures, resizing assets across twenty formats, drafting the fifth version of a headline for A/B testing - this is exactly the kind of repetitive, low-stakes execution that used to burn out talent for no creative reward. It's already the most common entry point for AI in marketing departments: 89% of B2B marketers who use AI say they use it for generating marketing copy, making it by far the most common application. Handing it to AI doesn't have to mean cutting corners – a human still has to review everything at the end of the day.
At its core, AI was created with the intention of being an execution machine. Repetition, sometimes precision, speed and efficiency is what AI best offers us. Mundane tasks, when taken over by AI, can truly help condense lead times and leave creatives the room and capacity to create work that really matters.
Where AI needs to stay in its lane
None of the above means AI should be steering the ship. The moment it starts generating the strategy instead of executing it, people can tell.
As much as you think it might, AI really doesn't know your customer. It knows patterns that resemble your customer, pulled from an ocean of data that includes your competitors, their competitors, and roughly everyone else on the internet. It has no proximity to the actual anxieties, contradictions, and even inside jokes of the people your brand is trying to reach. That proximity is not a technical problem AI will eventually solve with a bigger model. It's a human problem, and always will be.
This is why the brands leaning on AI for strategy, not just execution, tend to produce work that feels weirdly familiar the moment you see it. Competent, brand-safe, occasionally even quite polished. But also utterly forgettable, because it was optimised for pattern-matching instead of built from an actual point of view and authentic angles built from the foundation of studying and living among audiences.
The gap isn't just a hunch either: in a controlled experiment by the Nuremberg Institute for Market Decisions, the exact same ad was rated as less appealing and less emotionally resonant the moment participants were told it was AI-generated rather than human-made.
The integration that works must look less like replacement and more like division of labour. AI can handle volume, speed and scale. Humans handle judgement: what's worth saying, who it's really for, and whether it's actually true to the brand or just true to the data published and scraped from the internet.
Integration should be supplementary (definitely not a shortcut)
The mistake most marketing teams make with AI isn't using it too much or too little. It's using it without a regimented system. Someone on the team discovers a tool, starts running briefs through it because it's fast, and a few mediocre campaigns later nobody can quite explain why every piece of content sounds boring, feels empty and hollow, or occasionally off-brand.
Real integration means deciding, intentionally, where AI can sit in the workflow and where it shouldn't. It means briefs still get written by people who understand the business, not generated wholesale and hastily approved by management. It means AI output gets reviewed by someone with taste and in tune with what's relevant before it goes anywhere near a customer, not published because the algorithm said it should perform well entirely because of baseless historical patterns. It means measuring not just output volume and metrics but whether the work is landing, building communities, and being remembered.
Done properly, this isn't marketing getting smaller, sloppier or lazier. It's marketing moving faster at the parts that are mechanical, so more energy is poured into the parts that matter.
The brands that get this right won't look like they're using AI at all
When integration is done well, it's invisible. Nobody looking at a genuinely great campaign is thinking about which parts of it were AI-assisted, because the thing that made it great, the insight, the timing, the nerve to say something real, was never something AI could have produced on its own anyway.
With AI becoming increasingly mainstream and normalised, we often hear "I bet this was made with AI" from virtually anyone on the streets chancing upon an ad or campaign. Public scrutiny is at an all time high during this time: half of consumers can already correctly pick out AI-written copy from human-written copy in blind tests, and the same research found that once people know they're looking at AI-generated content, over half say they feel noticeably less engaged with it. This makes prudent usage of these tools all the more cardinal in our processes as an industry.
The brands still arguing about whether to use AI or not are asking the wrong questions and whittling away the time they could be spending on the right things. The only real question left is whether you're disciplined enough to use it well - fast where speed matters, and nowhere near the parts of the work that only a person paying real attention could ever get right.