The Most Important Design Work Is the Work That Never Ships.

And AI is finally making it possible to do more of it.

There's a type of design work that energizes teams like nothing else. Greenfield. No constraints yet. The problem isn't fully defined, the solution space is wide open, and the work you make has the potential to change what your company decides to build next.

Speculative design. Vision work. The "what if we could do this?" work.

It's also, in my experience, the hardest work to resource.

When your team is heads-down shipping production work — and whose team isn't — spec work feels like a luxury. It competes for the same designers, the same time, the same sprint capacity. And production always wins because production has deadlines. Spec work doesn't. So it happens in bursts, when someone makes space, when a stakeholder gets excited, when there's a gap between projects. It's not a practice. It's an accident.

The cost of that is real. Spec work is how design gets ahead of decisions instead of reacting to them. It's how you shift an investment conversation, install a vision, make a future state credible enough to fund. When you can't resource it consistently, design becomes execution. You're building what you're told to build, not shaping what gets built.

I've felt this tension directly. Some of the highest-leverage work my team has done has been speculative — vision artifacts that changed how leadership thought about a product direction, prototypes that redirected roadmap conversations, explorations that surfaced architectural questions before they became expensive problems. That work mattered. It also required carving out capacity that was genuinely hard to justify when everyone was busy.

Here's what's changing: AI compresses the cost of speculative work dramatically.

The bar for spec work has never actually been low fidelity. Executives don't respond to wireframes. They respond to things that look and feel real — high-fidelity enough to be credible, detailed enough to provoke a real reaction. Meeting that bar traditionally meant pulling designers off production work for weeks.

AI changes that equation. A design leader with strong judgment and the right AI tooling can produce executive-credible vision artifacts faster and with significantly less team dependency than before. The thinking, the framing, the scenario design, the judgment about what's worth exploring — that's still human work. The artifact generation gets faster.

Which means spec work stops being a luxury and starts being a practice.

That's the shift I'm most interested in. Not "AI makes designers faster" — that's true but not that interesting. The more interesting version is: AI makes a category of high-value work accessible that most teams have never been able to sustain consistently. Design leaders who figure out how to use it that way aren't just working more efficiently. They're unlocking leverage their teams couldn't maintain before.

One question I'm sitting with: spec work was also where designers got energized. The greenfield, the creative range, the "what if" energy — that's not nothing. If AI absorbs more of that work, what replaces it for the team? I don't have a clean answer. But I think it's worth asking out loud before we assume the efficiency gains are pure upside.