AI is easy. The human part is hard.
Ross Gales
Sept 7, 2026
You can build an extraordinary AI experience that nobody wants to use.
As the technology gets better, building with AI is becoming increasingly accessible. Models are more capable, tools are easier to use, and new interfaces and agents can be created incredibly quickly.
But making an AI experience that people actually value is much harder.
Because customers don't experience the model. They experience the decisions we make around it: what it knows about them, when it asks a question, when it makes a recommendation, how much it explains, and when it takes action on their behalf.
And perhaps most importantly, when it gets out of the way.
These are fundamentally human questions, and answering them requires a much deeper understanding of the people we're designing for.
AI changes the relationship with the customer.
For the last couple of decades we've designed digital experiences largely around interfaces. Menus, search boxes, forms, filters and buttons, assembled into carefully considered journeys that help customers navigate from A to B.
AI changes that model.
An experience can now understand intent, interpret context, ask questions, make recommendations and increasingly take action on someone's behalf. That creates an opportunity to rethink how people interact with organisations entirely, rather than simply making today's interfaces a little smarter.
But it also introduces a different set of expectations.
If an AI recommends a car, a financial product, a healthcare pathway or somewhere to travel, people may want to understand why. If it knows something about them, they may want to understand how. And as we give these systems greater agency to act on our behalf, questions of permission, transparency and control become much more important.
The more capable these experiences become, the more carefully we need to design the relationship between the customer and the technology.
Trust becomes part of the experience.
We often talk about AI governance as something that happens behind the scenes. For customers, though, governance is something they experience.
It's in whether an answer feels credible and whether its limitations are clear. It's in how uncertainty is communicated, whether people retain control at the moments that matter, and whether there's an obvious path to a human when they need one.
None of these things feel particularly technical, but together they can determine whether someone embraces an AI experience or abandons it.
And the stakes extend beyond the interface, because customers aren't really deciding whether they trust your AI. They're deciding whether they trust you.
The technology won’t be the differentiator for long.
Most organisations will increasingly have access to similar models, platforms and capabilities. The technology underneath many AI experiences will become less distinctive.
What won't be commoditised so easily is a deep understanding of your customers.
Their needs, behaviours, anxieties and expectations. The language they use. The context they're operating in. The moments where they need reassurance and explanation, and the moments where they simply want something done.
Understanding those things is what allows us to make better decisions about where AI belongs in an experience and, equally importantly, where it doesn't.
That's why research, strategy, service design, experience design, content and testing become even more important in an AI-enabled world. They're not supporting disciplines around the technology. They're how we turn increasingly accessible technology into experiences people understand, trust and actually want to use.
And that's the opportunity I find most interesting about AI.
Not simply automating the experiences we have today, but designing entirely new ways for people to interact with organisations, access services, make decisions and get things done.
The technology might be getting easier.
The human part isn't.