March 9, 2024

Service Design and AI

What can service design teach us about AI? Our days are filled will patterns that we perform mindlessly. In fact, we’re constantly prompting ourselves to perform actions at home and at work.

We rarely stop to evaluate our own system. We don’t treat ourselves as our user, the despite the fact that we almost entirely depend on our internal “services”

Take a moment and think through your day. What are all the loops of behavior you’ve already performed without giving it much thought?
– Check the school lunch schedule and if it’s anything other that pizza make food for the kids unless it’s Tuesday or you’re out of bread
– Go through your list of tasks you have to complete that day and make room for them in the calendar, rescheduling or re-tasking as needed
– Review your email and prioritize those that are most urgent or relevant to your current focus. Reply to any from the sender “mom”
….. etc

These are services, backstage work that needs to *just happen* so you can get the stuff done that you want to focus on

Now, if you were to program an agent to perform these for you, how would you define a prompt to improve these services? What data would you automatically have it reference. How would you define the outcome you’re seeking? What constraints do you have to consider?

Now do that same exercise for yourself at work. Or, think about your customer and what you know about THEIR day and THEIR loops at home or at work

Same exercise – how would you prompt that? What other data or context is available for you to reference? How would you pull that in? What other services would you connect to? What would step one look like? What permutations do you need to consider?

You see what I’m getting at?

AI uses this type of logic to perform a task. It can loop those tasks. It can learn and improve those tasks. In many cases, it’s applied as a service

Service design, product design, product strategy – all of these factor in to this exercise. It’s a great way to start to think about applying AI in practice