'We absolutely want to integrate AI into our products,' clients say. As a digital agency, our response is: 'Hold on a moment.'
A great many companies currently want and demand the integration of AI into their digital products and services. As technology pioneers, we find this ambition exciting – but through our design lens, we see it positively while remaining somewhat critical, because in our view, value only emerges where technology and design complement each other meaningfully.
When it comes to integrating AI into products, this means: technology enables the use of AI, but design determines where it actually serves user needs and business goals. Neither creates value without the other – not for users, nor for businesses – and without both, AI becomes an end in itself. Which brings us to the first key insight from the talk:
The same principle applies to AI integration: its use must be a design decision that delivers value for users and businesses alike.
#1: Using artificial intelligence alone does not make a product.
The great opportunity for using AI arises where we find answers to two central questions: exactly what AI can contribute to a product, and what (on the business side) is needed to deploy it meaningfully. If we manage to develop solid answers to these questions, the door is open for AI to deliver genuine value for users and businesses alike. And this is where the second insight emerges directly:
'Magic' can already emerge from small applications like this alt text generator. Source: Storyblok
#2: Start with small, tangible solutions – e.g. mini AI assistants – and develop AI iteratively.
Ambition drives outstanding digital products. But companies and agencies must stay sensitive to when healthy motivation tips into counterproductive over-eagerness. Because visible impact doesn't require all-encompassing super-intelligence. 'Magic' often emerges from small things – from the well-considered, deliberately bounded use of AI that makes users' lives or work meaningfully easier at key moments. Positive side effects: the scope within which AI can operate stays manageable. We control which data we use to train AI until it meets our standards. And ultimately, the quality of AI output remains controllable – since it is inextricably linked to the input that preceded it. Which brings us to the third point.
The fundamentals of good dialogue design remain relevant when interacting with AI.
#3: AI assistants are not a magic text field – they are a dialogue that needs to be designed.
Looking at common patterns for interacting with AI, it's clear that input fields currently dominate – sometimes paired with additional controls – while new patterns are only just being discovered. As long as that's the case, one thing holds: designing for AI is essentially 'next-level dialogue design', where it is crucial that the AI interaction explains itself sufficiently.
The fundamentals of good dialogue design – as defined, for example, by ISO 9241 – retain their relevance despite, or because of, AI. The human-AI dialogue must be designed reliably for users – framing, transparency and control are the central areas of action in good interaction design. Users must understand what AI can be used for and what interaction options exist. They must always know what AI is doing and why. And they must be able to intervene or take full control at any time.
But how can designers take responsibility when designing with AI?
To integrate artificial intelligence responsibly, we should develop its value iteratively and make it understandable.
#4: Designers should approach AI as they would any technological innovation – with an alert mind and a solid methodology.
Looking at the enormous technological developments of the past ten, twenty or thirty years, it becomes clear that designers had to practise working with new technologies long before ChatGPT in order to deploy them meaningfully in the interest of users. Adapting to new technologies and actively exploring them has always been part of the craft.
Especially in times of technological disruption, the fundamentals that enable designers to make well-grounded decisions gain in importance. Well-trained designers already carry the necessary tools to develop solutions with AI and help shape the digital future.
The prerequisite for this, however, is that they actively work with AI. Only then can they truly understand how it works, its potential and its limits. It is only this understanding that enables them to challenge the status quo in the interest of genuine value.
AI is here to stay. And with it comes our responsibility: not to stand on the sidelines in amazement as AI is developed and integrated, but to become active – to try new paths and consciously shape our digital future.
Exactly what designers – and we at 21TORR – have always done. Just now with AI.
Many thanks to our Design Lead UX Nele Zschiesche for her valuable talk at the World Usability Day 2025.
We challenge the now to create the new.
Since 1994, 21TORR has been developing digital products and services together with its clients at the intersection of design and technology – with measurable business impact and genuine value for brands and people.
Let's create your AI future.
