AI+Design for imagining and designing new experiences and services

In autumn 2019, we brought Josh Clark to Italy for a workshop called "Designing for what's next" - a topic that is still very relevant today. The focus was on new types of interaction between people and interfaces, from conversational systems to AI.
At the time, discussing AI meant mainly talking about predictive algorithms and computer vision. The case studies we analysed that day revolved around recommendation engines, smart defaults - default interactions shaped by data, statistics and user preferences - and image recognition.

From this workshop, I'd like to highlight three key concepts that defined "what's next" at the time:

• interfaces were beginning to change and evolve, incorporating intelligent elements and responding to data and input beyond forms and clicks (or taps),
• as Josh put it, AI was becoming a new design material,
• a critical and thoughtful approach would be critical to designing automation and intelligence while maintaining utility, a sense of control, trust and safety for users.

Fast forward to today - what was next has become now, and these three seeds, which may have once seemed distant or futuristic, are now fully sprouted and ready to flourish.

At Tangible, these are issues we have been discussing and thinking about for a long time. They have guided a number of strategic decisions, from internal training to R&D, to refine our capabilities and evolve parts of our workflow.
I'm not talking about internal AI applications to optimise tasks or speed up deliveries - although we're exploring these with great care to maximise value. We've discussed this at design events and will continue to do so.
Rather, I'm talking about how we work with and for our customers - how we harness the potential of AI and help them develop their digital products and services in this direction.

If the way people interact with systems is changing, how can we rethink the way they explore and use digital products and services?

Designing new interactions

GenAI tools familiarize us with conversational interactions - mainly text or voice-based chat. Or, alternatively, by feeding them an image or piece of content.

How can we imagine new interaction models for our product or service through AI?
How can we provide new ways to query and browse data, supporting natural language and human questions instead of a multitude of fields and filters?
How can we design phygital products that respond to input from the environment?

For example, consider redesigning the customer support area of a B2C website - moving from a traditional support model to a conversational one, like an Ask Anything interface. This allows users to ask questions in natural language and receive immediate, personalised responses.

Designing dynamic, tailored interfaces

Spoiler: turning everything into chat is not the answer. The recent demo of Gemini made this very clear.
Some of the proofs of concept and products we've seen in recent months show a stark separation between the AI-powered part - typically a text-based chat - and the rest of the experience, which remains visually rich and click/tap-based.

How can we integrate the capabilities of AI with the state of the art in digital interfaces?
How can we deliver personalised outputs that are still highly interactive and visually integrated into the broader digital ecosystem?
How can we leverage years of investment in design systems - built by many companies - to enable dynamic interfaces?

F̶a̶i̶l̶ Test fast, test early

We're back in an era of rapid experimentation and high velocity. Some of the principles that shaped the last wave of innovation - social media, user-generated content, SaaS - are proving useful again.

How can we validate an idea quickly without incurring high costs?
How can we mitigate the risk of investing in an AI capability that may lack real value or adoption?
How does AI fit with existing business processes?

Problem framing and rapid prototyping are critical tools here. And thanks to AI, the ability to quickly create realistic prototypes has increased. It also enables multiple iterations, allowing us to explore different scenarios and hypotheses, enriching the learning phase.

Designing for trust

We've discussed and written a lot about this at Tangible - it's something we're passionate about. So I won't go into too much detail here.

How can we make our product transparent, clear and trustworthy?
How and when should we grab the user's attention and give them control, using mechanisms like mindful friction?
How do we communicate errors, approximations and the inherent "guessing" involved in AI-driven results?

We aim for a realistic and balanced approach because, as with any technology, good design is the key difference between understanding and rejection, utility and futility, effectiveness and ephemerality.

Good design is honest.
It does not make a product seem more innovative, powerful or valuable than it really is. It does not try to manipulate the consumer with promises that cannot be kept.
- Dieter Rams

So, what’s next?

This post is intentionally full of questions rather than answers.
In innovation projects, the phase of problem identification — or better yet, opportunity discovery—is where the dialogue between business and design has the most value today.

At Tangible, we have an R&D project dedicated to AI+design and the four points mentioned above.
For over a year, we’ve been inviting data scientists and AI experts to exchange ideas and broaden our perspective.

This week, we’re taking another step forward—spending two days working with Josh Clark on Sentient Design: an opportunity to explore how AI can make systems not only more advanced but also more context-aware and responsive to human experience.

Want to build the answers together?

We’ve created a workshop to turn uncertainties into concrete strategies. If you’re also exploring AI’s role in your product or service, let’s talk.