AI won't design your app for you.

Strategy · 6 min read

I've had several people come to me after spending weeks with AI app builders. They described their idea, the tool generated screens, they tweaked and iterated, and what they ended up with looked like an app. On the surface. Then they showed it to someone who works in their industry and the feedback was always the same: "This doesn't actually work the way our job works."

AI tools are getting genuinely impressive. I use them in my own work. They're fast, they're helpful, and they're changing how software gets built. But there's a gap between generating something that looks right and designing something that works right. And that gap is where most app ideas live or die.

Where AI helps

AI is brilliant at the mechanical parts of app building. Generating boilerplate code. Creating basic screen layouts. Writing first drafts of content. Suggesting colour palettes. Speeding up tasks that used to take hours. For a developer, AI tools can cut development time significantly on the repetitive, well-defined parts of the build.

For ideation, AI is useful too. It can help you explore different approaches, generate variations, and think through scenarios you might not have considered. It's a thinking partner. A fast one. And when you know exactly what you want, it can help you get there quicker.

Where AI falls apart

AI doesn't understand your users. It doesn't know that the person using your app has limited fine motor control and needs bigger buttons. It doesn't know that your industry has compliance requirements that affect every screen. It doesn't know that your target user is a fifty-year-old tradie who's never used an app like this before and will bail if the first screen is confusing.

Good app design isn't about generating screens. It's about understanding the people who will use those screens. Their limitations. Their context. Their frustrations. Their workflow. That understanding comes from research, from conversations, from sitting with the subject matter expert and asking "what happens when this goes wrong?" AI can't do that. It can only pattern-match against what it's seen before.

And then there's the complexity wall. AI handles simple, well-defined tasks well. But apps aren't simple. They have edge cases. Permission layers. Data relationships. Flows that branch based on user type. The moment your app has more than one type of user, more than one set of rules, or more than one way to complete a task, AI-generated output starts breaking down. Not obviously. Subtly. In ways you won't notice until real people try to use it.

The right way to use AI in your app project

Use AI as a tool, not as a replacement for the process. It's excellent for speeding up parts of design and development. It's terrible at replacing the strategic thinking, the user research, and the domain-specific decisions that make an app actually useful.

The best outcomes I've seen are when AI assists a skilled team, not when it replaces one. A designer who uses AI to speed up layout exploration. A developer who uses AI to write boilerplate faster. A founder who uses AI to draft content that a human then refines. That's where the value is. AI as an accelerator, not a substitute.

If you've tried the AI-first approach and hit a wall, that's normal. It doesn't mean your idea is bad. It means your idea is complex enough that it needs human thinking. That's a good sign. The simple ideas are the ones AI handles fine. The valuable ones are the ones that need more.

Sources
AI in UX Design (Nielsen Norman Group) - How AI augments but doesn't replace human-centred design.
AI in Design (Interaction Design Foundation) - The role and limitations of AI in the design process.

Related blog posts:

The difference between a feature and a solution →

Why subject matter experts build the best apps →

The recording problem →