I built an AI service mapping tool

Ok that was a bit clickbait.

Now, before you come at me for it, I will be the first person to list the problems with my own work.

This is really an experiment and a way for me to delve more deeply into the materials, logics and issues with AI assisted work, in this case, design processes.

I make things in order to think and learn. Building this has given me the chance to really interrogate, in code and language, what is going on.

As someone who talks and teaches so often that mapping services is about the act of collaboration, and knowledge sharing to identify a number of useful insights I’m not waving a flag here to say ‘we can make your map making more efficient’.

And AI assisted tools in design can be quite problematic in the sense that the de-labourisation at play here is for me, really, an antithesis of what design is and should be. We make to think, taking away making, for me means, taking away time to think individually and together.

It’s concerned me. I’ve observed a slow decline in the space for critical thinking over the past decade of the design industry with the over productisation of tools, in pursuit of the now expected ‘deliverables’

That’s not a critique of individuals, rather an observation of a system that has productised design practice into deliverables, outputs and rehearsed processes. It’s a by-product really of risk averse organisations, usually because design has been put into a square delivery bucket for digital things, or bad ideas, not a strategic flex. (This is not everywhere, but a pattern I’ve observed over and over)

So why experiment with this?

A service mapping tool in a browser asking a user to create an AI assissted

I’ve been teaching service design for years, as an educator, and through 100s of projects in my past life of Snook where we’d skill up our keen clients in design whilst we did the work.

I’ve observed more and more, an obsession with mapping and questions about ‘how to do it right’.

My answer is always – there is no right way, and no, the map is not your job’s output. It’s a process. Mapping, the conscious act of it, whether individually or in a group is a form of thinking through making, which for me, enhances verbal dialogue and prompts us to think differently, recall new information.

And you need to find the best way to communicate and tell stories that will help bring people with you and where you want to go. The map rarely does that, unless you work it up into a useful tool for collaboration or into a story based artefact.

So I didn’t build ‘an AI mapping tool’ per se

It’s a digital version of a tool we use at The School of Good Services in all our training, getting people to see services and map how they work for users.

I have found where people got really stuck is service blueprinting and trying to think through how they can map multi-channel experiences (which services often are) together.

So the logic I’ve built for this is to help people get started from a blank page. The video shows the simpler AI assisted mapping tool for singular user journeys.

If we ever did release this (although likely just to be in our training) it comes with every caveat I usually say which is ‘never make a map up’, only if you’re starting with a clear assumptions exercise to be tested against research, or greenfield designing (rare).

So some constraints I put in to this;

It won’t fill it all in, it makes a start for you, you can continue to edit it
Always show if something is AI-assisted
It shows you how you map one vertical stage, to then build left and right from
It doesn’t lead too much, is poised to ask open questions based on what you’ve said
It reflects that you’ve said to then build
Everything is editable once it’s mapped
The service blueprint AI assisted feature listens to you and asks prompting questions to find out if the user can use another channel to do the same thing
I made the user activity in service blueprint ‘channel agnostic’ which to me, is key for mapping a multi-channel service

I learned lots

Making it really helped me to extract my knowledge, guidance and pointers I share with people on service mapping whether user journeys or blueprints, baking this into the conversational AI.

It was a helpful exercise, and what I’m excited about, is exploring AI’s capabilities to build creative tools bespoke to a knowledge base or particular task, rather than sloppy creative artifacts.

What’s next?

Keep refining the questions that the AI assisted mapping and conversation works

Build new ways for maps to be made that feel less linear and more emblematic of relational services

Obvious UI/Accessibility work to be done and testing

Think about use cases of where I’d use this in training/other scenarios and consider the user flow and guardrails that comes with it in terms of ensuring you’ve done your research