How I Use AI Day to Day as a Designer
The parts worth keeping after the novelty fades
There is a deafening amount of noise right now about AI replacing designers.
If you spend too much time on Twitter/X, you’d think Gemini and ChatGPT have already rendered us obsolete.
I’ve been doing this for over 25 years.
I’ve worked at EA, Disney, Nintendo, and Warner Brothers. I’ve seen tools and trends come and go—Flash, Skeuomorphism, the rise of Mobile, the shift to Figma.
Every time a new technology drops, the panic sets in. And every time, the job doesn’t disappear; it just evolves.
I can tell you with absolute certainty: AI is not going to replace the strategic, empathetic, human work we do. It cannot replicate the nuance of negotiating with a stakeholder, or the gut feeling that a UI flow “just feels off.”
However, it has completely changed how I handle the parts of the job I hate.
The “Executive Function” Tax
I was diagnosed with ADHD later in life, at 47.
Suddenly, my entire career made sense.
It explained why I could hyperfocus on a complex design system for 12 hours straight, losing all track of time, but couldn’t write a simple project brief to save my life.
For neurodivergent brains (and honestly, for many neurotypical ones too), the hardest part of design isn’t the designing. It’s the starting.
It’s the “Executive Function Tax.”
It’s the energy required to switch contexts from “big picture thinking” to “writing an email.” It’s the paralysis of the blank canvas. It’s the administrative friction that sits between having an idea and executing it.
I don’t use AI to “design” for me. I don’t ask it to “make me a website.” That results in generic garbage.
Instead, I treat AI as a tireless, highly efficient Junior Designer who sits next to me.
This Junior Designer doesn’t have great taste, but they are incredibly fast, they never complain about revisions, and they are excellent at getting the ball rolling so I can take over.
Here are three specific ways I use this “Junior Designer” in my day-to-day workflow to bypass my brain’s roadblocks.
1. The “Blank Page” Destroyer (User Research)
One of the biggest hurdles in any project is the initial research phase. You have a vague brief, and you need to start understanding the user. Staring at a blank Notion page trying to conjure “User Personas” from thin air is a recipe for procrastination.
I used to put this off for days. Now, it takes 15 minutes.
I paste the project brief into my AI tool of choice (usually Claude or ChatGPT) and run a prompt to generate a “Strawman Persona.”
Based on this project brief, generate 3 detailed user personas. Include their frustrations with current solutions, their technical proficiency, and 5 specific questions they would ask during an onboarding flow.
The Result: The output is usually about 60% right. It’s often a bit cliché. But that doesn’t matter.
Because it is infinitely easier to edit a bad persona than to write one from scratch.
I can look at the output and say, “No, that’s wrong, our user isn’t tech-savvy,” and immediately correct it. The AI breaks the seal on the blank page, allowing my actual expertise to kick in.
2. The “Unstuck” Generator (Microcopy)
Nothing kills my flow faster than stopping high-fidelity UI work to figure out the text.
You know the feeling: You’re deep in Figma, designing a complex checkout flow. You’re moving pixels, balancing hierarchy, checking contrast. Then you hit an empty state or an error message.
Suddenly, you have to stop being a Visual Designer and start being a Copywriter.
“What should this error say?”*
“Is ‘Whoops’ too casual?”
“Is ‘System Failure 404’ too robotic?”
I used to stare at the cursor for 20 minutes, letting all my visual design momentum die.
Now, I use a prompt to generate 10 tonal variations instantly.
Give me 5 options for a ‘Credit Card Declined’ error message. Tone: Empathetic but clear. No technical jargon. Keep it under 140 characters.
I rarely use the exact output it gives me word-for-word.
But seeing 5 decent options instantly gives me the momentum to combine them into one good solution and move on. It turns a 20-minute blockage into a 2-minute task, keeping me in my flow state.
3. The “Brutal Critique” Partner (QA & Accessibility)
We all get “designer blindness.”
After you’ve been staring at the same set of screens for 6 hours, you stop seeing the mistakes. You miss contrast issues. You miss logical gaps. You forget that users don’t know the product as well as you do.
In the past, I’d find these mistakes two days later during a stakeholder review, which is... painful.
Now, before I show work to anyone, I paste screenshots or copy descriptions into AI and run a specific critique prompt.
Act as a senior UX researcher. Critique this user flow based on Nielsen’s 10 Usability Heuristics. Be ruthless about potential accessibility failures and visual hierarchy issues.
It acts as a fresh pair of eyes. It will point out things like:
“The ‘Cancel’ button is too close to the ‘Confirm’ button, which violates error prevention heuristics.”
“The grey text on the grey background likely fails WCAG AA standards.”
It catches the silly mistakes so I don’t have to. It allows me to walk into meetings with confidence, knowing I’ve already stress-tested the logic.
Building a System, Not Just Prompts
The key to making this work isn’t memorizing a thousand random “magic words.”
It’s about integration.
Over the last 18 months, I realized I was using the same scripts over and over again. I stopped trying to be a “Prompt Engineer” and started treating these prompts like any other design tool—like a Figma component or a sketchbook.
I started keeping a personal library of the prompts that actually worked reliably in production environments.
These aren’t “make a picture of a cat in space” prompts. They are “write a usability script for a 30-minute interview” prompts.
I realized this library was saving me about 10 hours of busy work a week.
That’s 10 hours I can spend actually designing, or spending time with my daughter, or just not feeling overwhelmed.
The Designer’s AI Toolkit
I recently cleaned up that personal library. I organized it, added my notes on why they work, and put it all into a Notion dashboard.
I’ve packaged it up as The Designer’s AI Toolkit.
It includes the 30 specific prompts I use for:
UX Research (Personas, Journey Maps)
UI Ideation (Color palettes, Layout ideas)
Content Generation (Microcopy, CTAs)
Feedback (Heuristic analysis, Accessibility checks)
I also included my “AI Output Critic” system, which is a checklist I use to make sure the AI isn’t hallucinating or giving me bad data.
I put a $14 price tag on it. That’s basically the cost of a sandwich and a coffee.
I’m not trying to get rich off this.
I’m charging a small amount because I spent real time curating it, and I believe that when you pay for a tool, you’re more likely to actually use it.
If you’re a designer who feels overwhelmed by the “grunt work,” or if you’re neurodivergent and tired of fighting your own brain to get administrative tasks done, I think this will help you.
If you want to grab a copy, you can download it here.
P.S. If this helped, I’d really appreciate a restack.
It helps the piece travel and keeps me motivated to keep writing practical stuff for designers navigating the AI era.