Design isn't the deliverable: It's the hidden work that makes products succeed

You shipped a new feature. The screens look great. The interactions are smooth. But conversion didn't move.

The problem wasn't the design. The design was beautiful.

The problem was everything that should have happened before the design. The discovery that didn't go deep enough. The assumptions that weren't tested. The stakeholder alignment that never fully happened.

The deliverable was fine. The invisible work was missing.

What AI can see vs. what it can't

AI is exceptional at the visible parts of design.

Give it a prompt, get a landing page. Describe a style, get illustrations. Explain your brand, get a logo. The outputs are often good, sometimes surprisingly good.

But AI can only work with what you give it. And the hardest parts of design are figuring out what to give it.

What problem are we solving? For whom? What matters most? What constraints are real and which are assumed? What's the one thing this design needs to accomplish?

These aren't prompts. They're discoveries. And discovering them requires intuition, observation, and the kind of pattern recognition that only comes from years of watching products succeed and fail.

AI can help you explore once you know the direction. It can't tell you which direction to explore.

The gut check layer

There are always multiple ways to solve any design problem. Always.

AI makes this more visible than ever. Generate five approaches in an hour. See options you wouldn't have considered. Explore variations endlessly.

But more options don't mean better decisions. Often they mean harder decisions.

At some point, you have to choose. And choosing requires weighing factors that can't be fully articulated: Does this feel right for the brand? Will users understand this immediately? Does this solution scale to the next three features we're planning? Will stakeholders trust this direction?

I call this the gut check layer. It's not arbitrary, it's informed by experience, by context, by all the details you've absorbed about this specific project and problem. But it's also not reducible to logic.

AI can generate the options. Your gut—trained by years of doing this—picks the path.

“
AI can generate the options. Your gut, trained by years of doing this, picks the path.

People decide when they see

One of the most important things I've learned: people struggle to evaluate abstract ideas. They find it easier to evaluate concrete things. Something they can see and "touch".

Tell a stakeholder about your concept and they'll nod along. Show them a working prototype and suddenly they have opinions, concerns, ideas they couldn't articulate before.

This isn't a flaw. It's how human cognition works. We need to see something before we really know how we feel about it.

AI has transformed how fast we can create things worth seeing. In the past, you'd wireframe, then mock up, then maybe prototype—weeks of work before getting real feedback. Now you can go from concept to functional prototype in hours.

That's not about replacing the process. It's about accelerating learning.

The prototype isn't the deliverable. It's a tool for figuring out what the deliverable should be. A way to bring stakeholders on the journey. A mechanism for discovering what you didn't know you didn't know.

The faster you can show something real, the faster you can learn whether you're on the right track. AI makes that faster.

But knowing what to show, what to test, what questions to ask, that's still craft.

“
The prototype isn't the deliverable. It's a tool for figuring out what the deliverable should be.

The alignment loop

Design doesn't happen in isolation. It happens in collaboration, with founders, product managers, engineers, marketers.

Each of these people has context you don't have. Constraints you can't see. Priorities that shape what "success" means.

The work isn't just solving the problem. It's solving the problem in a way that everyone understands, believes in, and can execute.

This happens through iteration. Showing work early. Getting feedback. Adjusting. Showing again. Gradually building shared understanding of what we're making and why.

AI can speed up parts of this loop, generating the next version is faster than ever. But the loop itself? The process of alignment, of bringing people along, of building conviction? That's human work.

Skip it and you get designs that technically solve the problem but never ship. Or ship but get undermined. Or get built but don't perform because something was lost in translation.

Building brand through AI, carefully

Here's where it gets interesting.

I use AI to generate brand imagery. Blog graphics. Marketing assets. Visual elements that need to feel like they belong to a specific brand.

AI can produce these faster than any human. And sometimes the results are genuinely inspiring, combinations and styles you wouldn't have thought to try.

But consistency is the challenge. The same prompt yields different results. Models change. What worked last month doesn't work the same way today.

So the craft becomes codification. Building systems for getting repeatable results. Developing the judgment to recognize when an output is on, brand versus just close. Learning to guide AI toward a specific visual language and maintain it across months of assets.

AI accelerates production. But the creative direction, the taste, the coherence, the ability to build a visual system that holds together, that's human.

The 90% premium

Here's the economic reality:

The visible 10% of design—the screens, the assets, the deliverables—is getting commoditized. AI can produce it. Quickly. Cheaply. At quality levels that would have seemed impressive five years ago.

The invisible 90%—discovery, decisions, iteration, alignment—is not getting commoditized. It requires context, judgment, relationships, and the kind of thinking that can't be captured in a prompt.

When you hire a design team, you're hiring the 90%.

You're paying for the discovery that ensures you're solving the right problem. The decisions that shape the right solution. The iteration loops that build alignment. The judgment that makes AI-generated outputs coherent and effective.

The deliverable is what you see. The work is everything you don't.

And that's where the value lives.