the 5% AI can do

The reality of adapting AI tools in your design process

Sep 01, 2025

Hi, I’m Maureen. I work as a product designer and design educator. The Cursor Magazine is my online publication on design and career.

Let me start with a confession: I’ve been struggling to adapt AI technologies into my workflow in a way that actually saves time and increases efficiency and output quality.

Initially this struggle came from overwhelm. With so many tools hitting the market, which is should you try first? And what parts of my work are the ones that actually can be enhanced (or even replaced) by AI?

By trying out different AI tools, the overwhelm was replaced by a growing sense of skepticism. Aside of privacy and security concerns that makes it hard for in-house designers to just adapt AI tools into their workflows, I also noticed that it was really difficult to accurately replicate my designs in tools like Lovable. I may be saving time by not having to create designs from scratch in Figma, but I was spending that extra time on getting workable outputs in AI prototyping tools.

I started to doubt myself. Everyone is vibecoding, releasing products on ProductHunt and meanwhile I’m struggling to adapt AI tools into my workflow. What was it that everyone else understood but me?

I tried to do all my design work in AI tools. Guess what: it doesn’t work.

When I actually paid attention to how and when I was using AI in my workflow, I noticed that AI tools today only enhance the shallowest 5% of my work. They can help me with early wireframes, surface-level research summaries or quick concept sketches.

Frankly, I can’t wait for the day I don’t have to spend hours pushing pixels, fixing spacing or building repetitive flows. I want tools that can free me from that tedium so I can focus on the parts of design that bring me the most joy. Stakeholder alignment, conducting nuanced research, getting buy-in to change an experience in radical ways, balancing short-term wins with scalable solutions, listening to the conflicting wants of both business and customers and finding a compromise. There’s so much to design that can’t be solved with a simple prompt. Those parts that happen off-screen make up the vast majority of my job as a designer and can’t be solved with a simple prompt.

This moment in AI advancement leaves me torn. Part of me is disappointed because I hope for deep enhancement, not shallow shortcuts. I don’t want AI to just speed me up on the easy parts. I want it to meaningfully expand what I can do as a designer. Another part of me is relieved. If AI only skims the surface, then it can’t replace the craft, the judgment, the invisible labor that design really is. Maybe that’s the point.

With giving in to the speed and convenience that AI promises, we’re giving up on our own willingness to learn, to think and to stick with a task when it gets challenging. To tap into our own creativity and problem-solving skills and learn new skills.

Sitting with an idea and trying to figure it out myself isn’t idle time, but an important contributing factor to the quality of my design work. Employing AI tools would allow me to be faster, yes, but they would take away that valuable time to sit around, think and process.

And so, I wonder. Maybe it’s not AI that will take our job. Maybe it’s not another designer using AI who will take our job. Maybe it will be our own surrender to convenience and speed, trading creativity and critical thinking for reliance on AI, that ultimately undermines our careers.