Updating the Service Design Toolbox for AI-Powered Experiences

Pontus Wärnestål
4 min readMay 13, 2018

We live in a world under construction — on the brink of an infrastructure paradigm shift. Such shifts occur now and then throughout history, and they drastically change a large number of people’s reality, and society itself.

Industrial paradigm shifts. We’re now entering the fourth one, with new design opportunities and requirements.

As designers, we need to be able to quickly deploy into new territories, and address problems that our classical toolbox might not be optimized for. As most readers know, we live service-providing economy where 70–80% of the economy is comprised of services. And as the economical standard around the world increases, this will be true globally eventually.

The human capital (service-oriented wealth) is increasing globally. (Source)

For example: you no longer own the CD. Instead, services such as Spotify gives you access in the devices and places where you require the service: CarPlay, on your Phone, at home streaming through e.g. Echo or Sonos, or on your sports watch with an adaptive BPM on your music to match your running speed, etc. The service is “atomic”, meaning that it can be configured on multiple platforms, and is not tied to a specific physical medium.

Examples of Spotify devices and contexts.

Today, data-rich platforms and algorithms affect not only digital services, but creep into the physical spaces and infrastructure with societal implications. For example, a lot of people have replaced their ownership of a car with on-demand access. Uber and Lyft and other actors are doing to car ownership what Spotify did to CD ownership. And this affects infrastructure in terms of mobility and transport.

We’re at this tipping point because of cheap mobile computers, cloud services, Machine Learning API:s, and open data standards. 100’s of millions of people are using affordable tech such as phones to send all kinds of information to each other, at near-zero marginal cost. So, the internet is — obviously — no longer something that provides your browser access to media content; it instead unlocks real-world services.

So the next wave of services is increasingly AI-powered. And as more and more aspects of AI are getting commodotized, completely new kinds of services emerge. Such services allow us to outsource the service to the object itself — which then acts on our behalf. (This class of systems is called agentive technology.)

The shift of digital services into the physical world in the wake of data-driven AI, powered by tens of billions of sensors in 2020 (the estimates differ a little), means that not only software, but also ubiquitous technology, is quite literally eating the world.

This will (and does already) greatly affect the infrastructures on a global level. And with this type of design space, it’s not enough to have the “desire to build delightful things”. As designers we have a responsibility to humanize technology for positive societal impact.

If we don’t design our way through this paradigm shift carefully, we are in trouble. Because the famous “invisible hand” is also blind. Without purpose and meaning, we are lost.

That means we need to re-invent and re-combine our toolbox to (a) ask the right questions, and (b) cater for these new design spaces.

Asking the Right Questions

A good tool should encourage us to ask the right questions. With this human-centered design canvas for networked platform services, we can connect the notions of algorithm impact and enhance the resources and data that augment human workers, in order to create magic user experiences on a networked platform of services.

The Augmented Service Platform Canvas.

This tool is, just like many other “canvas tools” out there, for getting an overview; it needs to be accompanied by other more detailed maps and models when it comes to the specifics. It puts the ethics and the impact at both the beginning and end so that you can fill out the areas in any order. Hopefully, it will help you as a designer to map relationships between data (quality, coverage, types, etc.), algorithmic effects using the platform (including other systems and services) to facilitate a service ecology that gives rise to magical user/citizen/patient/student/human experiences.

The canvas is tool-making in progress, so as it’s being used, I’m counting on it being updated and modified.

The tool has been used in the AI.m program at incubator High5, connected to Halmstad University in Sweden, as well as in course projects. To date, several companies from different sectors have used the tool and report that they feel a lot more confident in their ability to reason strategically about the connection between data, AI, end-user experience, and how it will affect their business.



Pontus Wärnestål

Deputy Professor (PhD) at Halmstad University (Sweden). Father of two. I ride my bike to work.