Last week, I hosted a small cohort for The Workshop Workshop. Unsurprisingly, we talked a lot about whether AI and machine learning can help with workshop design.
For the most part, it is still too early. Not because AI can’t string together an agenda, but because the kind of knowledge workshop design depends on is not written down in a way AI can learn from.
Have you ever searched the web or tried to find a book about workshop design? There is not a lot out there, and definitely not enough about the design part of workshop design.
What’s Missing From ‘Workshop Design’ Content
Most of what you find falls into two buckets: workshop types and workshop activities. You get lists of “strategy workshops,” “retrospectives,” “design sprints” on one side, and “brainstorming,” “dot voting,” “journey mapping” on the other. Most resources jump straight to activities and skip the harder step of picking the right kind of workshop in the first place.
There is hardly any content at all about the design process. This is why I created The Workshop Workshop in the first place. Most of the content just tells you what types of workshops exist or what types of activities you can run. They rarely explain the process of designing a workshop: applying user experience design principles to design a session that will produce specific outputs and help move your work forward. They do not encourage you to think about your workshop participants as users you are designing for. In The Workshop Workshop, this is the muscle we practice: making deliberate choices about the kind of workshop you are running before you even touch the activities.
Why So Many Workshops Are Still Bad
And this is why there are so many bad workshops. A lot of people are not really designing sessions; they are stringing together familiar activities and hoping they add up to something useful. They are not thinking about the workshop type, the mechanics of the session, the experience of the participants, or the set up of the activities. They are not designing workshops with intention. And if they do, they are not writing about it.
What AI Can’t See About Your Session
Which is also why AI is no good at designing workshops. It reflects what it can see. If all it has is lists of types and lists of activities, it will give you some plausible mix of those. It skips choosing the workshop type. It does not assess whether the activity is the right one to generate the outputs you actually need, with the people you actually have, in the time you actually have.
It also does not consider the participant experience beyond the most generic level. It does not know that this team has never worked together before, that the topic is politically charged, or that the senior sponsor tends to hijack discussions at the twenty‑minute mark. It cannot imagine disaster scenarios and design around them. It can suggest methods it has seen before, but it cannot promise you that the session will achieve its goals.
And honestly, why would we want it to? Why would we want AI to take the most interesting part of the work? I would rather AI handle the boring stuff: sending invitations, tracking the attendee list, wrangling calendar collisions, maybe generating a rough set of slides I can critique. I like the creativity of figuring out the right fit for purpose. I like building an experience that people will enjoy and that will be productive.
Craft, taste, and design are human endeavours. They are why I do this work in the first place. I am sure AI will get better at this over time, but right now, workshop design still depends on judgment, responsibility and context in a way models cannot replicate. So while I know it is just a matter of time before AI can do more of the design work, I plan to keep enjoying the art and science of designing workshops myself for as long as I can.
