What if the most valuable thing you could design this week is something that's deliberately wrong?
We're trained to reduce ambiguity. Tighter flows, cleaner components, fewer open questions. But some of the most consequential design decisions happening right now — which futures to invest in, which technologies to bet on, how AI reshapes whole workflows — can't be resolved with a polished prototype. They need something rougher.
What does speculative design actually look like?
When people hear "speculative design" they picture gallery installations or design school thesis projects. Something conceptual. Impractical. But the most interesting version of it happening right now looks nothing like that.
It looks like a team sharing rough prototypes of features that might never ship. Collecting weak signals from the edges of technology. Building shared theories about where things are heading. And — this is the part I find most compelling — making possible futures tangible. Including the ones that go badly.
I've started thinking of speculative design as a form of fortune telling. Not because it's mystical, but because both practices are fundamentally about surfacing assumptions. You make a future concrete enough that people can react to it. Argue with it. Get uncomfortable. That reaction is the actual output.
Why this matters more than it sounds
Most organisations don't fail because they picked the wrong future. They fail because they never made any future concrete enough to properly evaluate.
You can see this pattern right now with AI adoption. The same mistakes that plagued digital transformation a decade ago — siloed thinking, no genuine leadership buy-in, poor strategic planning — are showing up all over again. And I think the root cause is the same each time: nobody could picture what "done" actually looked like, so everyone defaulted to their own private assumptions about what they were building toward.
Speculative design fixes this. Not by predicting the right answer, but by giving a team something specific to point at and say "yes, this" or "absolutely not."
The failure state nobody designed
Here's a concrete example of what happens when you skip this step.
LLM chat became the default interface pattern for AI almost overnight. It happened so fast that most teams shipped without asking basic questions about who might be excluded. Now we're discovering — after these interfaces are everywhere — that many of them are deeply inaccessible to people who rely on assistive technology. The problems are real, they're widespread, and they're genuinely hard to retrofit.
Could speculative design have caught this? Possibly. If someone had built a rough prototype asking "what if chat becomes the primary way people access information — who can't use it?" before the pattern became dominant, that conversation would have happened years earlier. Not through user testing on a finished product, but through a deliberately rough artefact designed to surface exactly this kind of blind spot.
That's the real difference between speculative design and regular prototyping. Regular prototyping asks "does this work?" Speculative design asks "what happens if we're wrong about who this is for?"
Why we don't do more of it
I don't see "speculative design" on many job descriptions. It's still treated as a luxury — something for well-funded research labs or innovation teams with no deadlines. But the version of it that actually works in practice is far less glamorous. It's a working team sharing rough ideas, building shared vocabulary about what might happen next, and using tangible artefacts to have better conversations about direction.
You don't need a research budget. You need a willingness to make things that are deliberately incomplete.
And here's where it gets hard. We're so conditioned to present polished work — to show confidence, to demonstrate craft — that putting something rough and speculative in front of stakeholders feels almost irresponsible. Like you're wasting their time with something that isn't finished.
But that discomfort is the signal. If the future you're showing people feels too comfortable, it's probably not speculative enough to be useful. The whole point is to provoke a reaction. If everyone nods politely, you've built a regular prototype and called it speculative.
So, practically
Pick something your team is uncertain about. Not the next sprint — something further out. An AI capability you're not sure whether to invest in. An interaction pattern that might or might not become standard. A future where your product doesn't exist the way it does today.
Now make it tangible. A rough mockup. A storyboard. A five-minute demo of something that might never ship. Make it concrete enough to argue about, and show it to someone who'll actually argue.
The conversation that follows is the design work. And it might be the most important design work you do all month.