From AI hype to operational reality: What marketers need to do next

Anastasia: I don't think the fatigue is with AI – I think it's with AI theatre. We've had customers and brands leave events early because the content was generally unhelpful; it's all the same buzzwordy things. The generous interpretation is that, for the first time since the digital revolution, no one actually knows what's going to happen, but we stand on public stages feeling compelled to say something smart, so we end up saying nothing of real substance. We've also outgrown analogies like 'AI is your co-pilot' – it's not humans versus AI, it's humans and AI. That challenge feels like it's behind us now.

Steve: I'm probably slightly more optimistic. I spent a lot of time just wandering around, and I was impressed by how much data now plays a part in the process – when I looked at the designs that won and how much attention they'd paid to the data elements, that got me genuinely excited, because that's the world we live in. People were asking questions for the first time about how to actually put this together, rather than buzzwords or cost-cutting, which is maybe what I heard last year. It was more, 'I get this now – how do I use you to create something better?' I think we're past that acceptance stage.

Anastasia: ROI is obviously very important, but people treat it as the next step after onboarding. The reality is the next step is readiness – is your organisation ready and capable to operate differently than it did before? The tech part, in most cases, is the easy bit; you buy the tech, and it does what it does. Getting an organisation of thousands of marketers to do something differently, that's the heart of it. That readiness is often something many leaders don't think about until the train's already barrelling down the tracks, when really everything should start with: what change do we want to see, and are teams motivated to make it?

Steve: Even as a smaller business, at Dragonfly AI we're implementing new tools and joining our data together, and you think, 'I've got this already, I can just produce this report' – then you find the data is just slightly wrong. It doesn't understand the context of your business. At scale, with big companies, it reminds me of the days of the unified database, where everyone said, 'We're going to build a unified database of all our data' – and put literally everything into it. Two years later, having spent six or seven million dollars, they're like, 'We haven't actually done anything with this.' It's quite hard to get all your data sorted if you're a really big business.

Steve: I think maybe we all, when these things first came out, thought they were going to produce amazing creative themselves. I think we're past that now – what we've realised is what they really did was automate, in a really advanced, incredible way, large amounts of creative. If you need creative on every channel at any time, you're going to need these tools to help you produce it, otherwise you won't reach people the way they want to see and consume it. I'm now really excited about what Dragonfly AI can do: make sure that when you bring that creative, it's seen, it makes an impact, and it's remembered.

Anastasia: I struggle with the language here, because the way people talk about it is: what separates us from these machines is 'the empathy' – the machines do everything up to a point, then we sprinkle some empathy fairy dust on the final thing. I don't think that's right. In our work as marketers, empathy comes from a deep understanding of the customer, not oversight at the end. If you're not talking to a customer to get that information, you need to remember the connectivity to your primary sources. That's your job – to understand the customer and give the LLM that context, otherwise it's just fluent and convincing, not tied to what the customer actually cares about.