Designing AI together: when service design meets data science
Notes from UX Talk #42 with Komebi Studio
A couple of weeks ago we were invited to the UX Talk by POLI.design’s Experience Design Academy in Milan to talk about design of AI-based experiences and our framework to approach it. We felt it was a nice occasion not just to present Triple-O, but to develop the discourse in conversation with another practitioner. So we thought of Davide Posillipo from Komebi Studio to bring the complementary perspective of a data scientist who’s been arriving at the same questions from the other side.
If you want the full recap, the Experience Design Academy published a detailed write-up covering the main points. Here we want to share what stayed with us after the conversation, and where things are heading.
Finding the same wavelength
Most AI conversations (not only in design) move fast around new tools, capabilities, anxieties about what’s changing, there’s a constant pressure to keep up and have an answer. What we liked especially about this talk was the opportunity to slow down and discuss things from very similar mindsets of a healthy/critical approach to what’s happening.
Davide comes from statistics and data engineering, with fifteen years of building ML solutions inside organizations, while we come from service design — on paper quite different worlds. But as we started preparing together, we realized we’d been circling the same core frustration, that the current AI discourse tends to collapse everything into the instant: the instant productivity gain, the instant prototype, the instant return on investment. And this compression leaves out almost everything that actually matters: people, processes, accumulated knowledge behind the disciplines we practice.
At one point during the talk Davide described what he called a kind of stoic resistance to the AI hype — not rejecting the technology, but pausing to ask whether adopting AI in a given situation is actually the virtuous choice, or whether we’re being swept along by a pressure that isn’t really ours. It resonated deeply with something we’ve been building into the Triple-O Framework from the beginning: the idea that the spectrum of possibilities starts with avoid, and that choosing not to use AI in a specific context is a legitimate, conscious design decision, not a failure of imagination. It sounds straightforward, but it’s one of the hardest things to argue for when the whole room has FOMO, and finding someone who approaches it with the same philosophical seriousness was one of the most rewarding parts of the evening.
The space in between
The talk also gave us a chance to explore something we’ve been thinking about increasingly, which is what happens in the space between the design side and the data side of AI projects.
Davide brought a story from a recent Komebi project where they assembled what they call a “data design unit” — designers and data engineers working together through every phase, including sitting in each other’s interviews with stakeholders. When they spoke to the company’s internal data team, the people who actually keep the data infrastructure running said that nobody had ever asked them how their work connects to the bigger organizational picture, and they were genuinely surprised to be part of the conversation at all.
We’ve seen this pattern in our own projects too. Some of the most valuable insights emerge when people who normally operate in separate tracks suddenly see how their work connects, and better management of data flows can have a lot of impact on that — it can reshape how an entire organization manages knowledge, how departments collaborate, even how they relate to external providers. Capturing and measuring that kind of systemic change is where things get particularly interesting for us, because it turns out the qualitative methods we’ve always used in service design might be exactly what’s needed to make those impacts visible. We’ll have more to say about this in our upcoming Touchpoint publication, stay tuned!
Depth over speed
The conversation naturally drifted toward the future — what skills designers will need, how much the profession is really changing — with concerns that AI evolution moves in unpredictable jumps rather than linear progression, and that predicting the next step feels almost impossible. We found ourselves sharing something we don’t talk about often enough. In 2019, when we started our first AI-based service design project, we were mapping scenarios for what these experiences might look like five years later. Nobody was talking about LLMs yet, nobody could have predicted the specific form GenAI would take, but many of the things we mapped then — proactive agents, personalised services, zero-interaction experiences — are recognisably what’s happening now.
That wasn’t because we had some special foresight, but just as always we did the slow work — deep research with extreme users, careful attention to long-term human needs alongside technological trajectories, workshops with sci-fi writers, the kind of thinking that resists the pull of the immediate. And that’s the practical argument for depth over speed: not that we should completely ignore what’s moving fast, but that investing in understanding context, people, and the deeper currents underneath the hype is what actually prepares you for a future you can’t fully predict. The tools and interfaces will change — we might not be designing in Figma in five years — but the capacity to observe, make strategic decisions, understand what people actually need will not expire.
Bringing back the depth — whether it’s the history of statistics, or design traditions, or even a bit of Kant’s “spirit of depth” — is not nostalgia but a way to actually expand what the technology makes possible.
What’s next
We’re both publishing in the upcoming issue of Touchpoint, the Service Design Network’s journal — our article on AI impact measurement through a case study, Komebi’s on a collaboration framework between design and data disciplines. The two pieces are independent but deeply complementary, and we’re looking forward to seeing them in conversation with each other.
Komebi also published a longer reflection on what they call the post-technical data practitioner — where the data profession needs to go when technical execution becomes a commodity. Worth reading if you work anywhere near data and AI implementation.
And we’re continuing to develop the Triple-O through our bootcamps and project work, where every engagement teaches us something new about where the framework holds and where it still needs to stretch. We’ll keep sharing as things evolve — replies and thoughts are always welcome!