AI design workflow
How to build your first AI design workflow as a designer in 2025 and beyond.
Hey there! This is a ✨ free edition of ADPList’s newsletter ✨ designed to make you a better designer and leader. Members get access to exceptional leaders' proven strategies, tactics, and wisdom. For more: Get free 1:1 advisory | Become an ADPList Ambassador | Become a sponsor | Submit a guest post
Today’s newsletter is brought to you by Atono.
Tired of your Figma design getting butchered by devs? 👀 And PMs rewriting your UX logic mid-sprint?
You’re not alone. Here's how top teams are finally staying aligned—without extra meetings or messy Slack threads. 👇
Designers,
When GPT was first introduced to me by my friend, I never thought I would rely on something like AI for my tasks. Fast forward to today, I use GPT-4o for almost everything.
I was on a call with my manager and I asked him for his honest opinion - if he thinks AI can displace designers. (It’s honestly all over the internet.) Not now, maybe but what about when we’re on GPT-7 or 8?
I already see designers using tools like Firefly, Runway, and Midjourney (at least some, if not all). According to a Weavely.ai survey, 86% of designers rely on ChatGPT for work.
But the real question is, if designers are using AI, then what is the best way to do that?
While scrolling through the internet, I came across multiple examples showing how just the right prompt & resources can make the work much easier.
Coming back to the question, he told me clearly that the problem isn’t with AI, it’s whether designers know how to best utilize it.
There’s no chance a designer can be replaced if they know how to make AI work for them.
It’s simple: “AI won’t replace you because someone still has to make it work.”
So let’s dive a bit deeper into this👇🏻
Are AI designers the big thing now?
I was scrolling through X to see how designers are actually utilizing AI in the best way possible. And designers casually post mind-blowing ways in which they use AI in their workflows. I’m not even talking about generic “use ChatGPT for ideas” posts. I’m talking about real, tangible applications, like someone building a Figma plugin that turns designs into production-ready code with Claude, or another using AI to test dozens of UX copy variations in seconds, tightening tone without losing brand voice.
I realized that the best designers, not necessarily the most experienced, but the most adaptive, are using AI like a second brain. Not to replace their creativity, but to stretch it further, faster.
Here are some of the best examples I saw:
These are just a few examples, but honestly, you can find hundreds like them if you spend just 10 minutes online. The point isn’t just to be impressed, it’s to get curious. What can you build with this? How can you integrate AI into your process in a way that doesn’t just save time, but sharpens your thinking?
1. How to speed up your ideas with AI?
Designers often start from a blank page.
I agree, ideation is supposed to be messy but it shouldn’t be slow. Whether you're building a new product, improving a flow, or designing a micro-feature, ik the blank canvas problem is real. And yes scrolling for ideas is so draining after a few minutes.
So, how do we use AI in the best way possible?
Let’s take a real-world example: a product designer working on a smart kitchen companion app. Instead of waiting for a team ideation sprint, they dropped a simple but structured prompt into ChatGPT:
“Give me 20 app feature ideas for a kitchen assistant app targeted at busy professionals. The app should use voice input and AI. Make them functional, not just fun.”
Within seconds, they had:
Meal planning suggestions based on time + ingredients
A real-time cooking assistant with voice-guided timers
Integration with health tracking apps
Auto-generated grocery lists with price comparison
A leftover management system for sustainability
Voice-based recipe shortcuts
…and 15+ more.
That’s not final design work. But it’s more than enough to spark team discussions, group ideas into themes, or sketch flows for concept testing.
(Btw, I was reading that a McKinsey study on AI productivity found that teams using generative AI in early-stage brainstorming reduced project timelines by up to 40% without compromising creativity.)
To make ideation with ChatGPT high-signal and useful, use this formula:
“Generate [#] design ideas for a [function] used by [user type], constrained by [platform/limitation/goal]. Ideas should be [tone: innovative/practical/emotionally driven].”
Examples:
“Give me 10 feature ideas for a journaling app for Gen Z that only uses voice input.”
“Suggest 15 onboarding concepts for a fintech app targeting first-time investors. Keep it mobile-first.”
“Brainstorm 10 AI-powered features for a productivity app designed for remote teams with 2-5 people.”
***
2. How to use AI in problem reframing?
If there's one thing I've noticed working in design and product teams, it's this: we're really good at jumping straight into wireframes, ideation, and solution mode. It's exciting. It's fast. And it feels productive.
But the biggest breakthroughs often come from how we define the problem in the first place. I know it sounds basic, but think about it: how often do we really sit with the problem before sketching the solution?
In a 2024 InVision survey, 78% of design leaders said misframing problems are the no. 1 cause of wasted design effort.
So here are some of the best prompts to use for problem-framing. Try them with your product/design brief or user issue:
What is the root cause of the problem and how can it be addressed?
How can you reframe the problem in a way that opens up new solutions?
How can you define the problem in a way that takes into account all relevant stakeholders?
What are the key drivers and factors contributing to the problem?
How can you define the problem using a systems perspective?
What are the potential unintended consequences of defining the problem in a certain way?
Here are some more prompts you can find to use:
3. Design research with AI
If you’re a designer in 2025, chances are you’re already using AI in some part of your research process. Maybe it’s to draft better interview questions, generate survey logic, or just to help structure your thinking when everything feels a bit foggy.
But let’s be honest, the real headache doesn’t begin until the interviews are done.
After you’ve run 10, 15, maybe even 20 user interviews… that’s when it hits you: now you need to actually analyze them.
So to make this easier, you can directly upload transcripts or interview notes.
Here are a few prompts you can use:
“Summarize common pain points mentioned across these interviews.”
“Cluster these responses into themes and suggest 3 personas based on behavior.”
Tools like Dovetail, Notably, and Grain now let you plug AI into video or text data for near-instant clustering of user sentiment.
It’s like having a fast-thinking co-researcher who never gets tired or overwhelmed. Once the data is summarized, use it to generate takeaways for the stakeholders:
“Write 3 key takeaways from this data that would be relevant for the product team.”
4. How to craft sharp UX microcopy
Now, let’s talk about the overlooked, under-optimized part of your design file: that tiny line of copy on a button, tooltip, modal, or form error. Because in high-pressure sprints or agency environments, designers are using AI tools to rewrite UI text without leaving Figma, and they’re shipping sharper, faster, and smarter.
While scrolling on X I came across a designer where she shared how she used a Figma plugin powered by ChatGPT to test 15 variations of CTA buttons for a retail app, each with different urgency levels and tones. They A/B tested five of them. The highest-performing one? AI-written.
Here are some great examples I found where AI has been used in one of the best ways.
The best part? No more placeholder lorem ipsums or last-minute scramble for “final copy.” If you’re not already using ChatGPT or Claude for UI text in Figma, you’re working 10x harder than you need to.
5. Getting design feedback
Design critique used to be a team sport: gather your Figma, call in a few eyes, brace for feedback.
But what if you could get meaningful, real-time critique from an AI before you even shared your mockups?
Welcome to the new wave of AI-powered design feedback, where tools like Visualizationary and custom GPT-4 workflows help novice designers level up, fast, and seasoned designers spot blind spots faster than ever.
How it works?
Let’s say you’ve designed a product analytics dashboard in Figma. Then you copy the Figma frame, plug it into an AI plugin or a vision-enabled GPT prompt, and ask:
“Can you give me visual hierarchy feedback and accessibility issues on this layout?”
Here’s what AI feedback might look like:
“The use of three shades of blue makes it hard to distinguish the primary CTA. Consider reducing the number of blue tones to increase visual hierarchy.”
“The chart legend is not accessible to colorblind users. Use patterns or labels instead of color alone.”
“Your ‘Upgrade’ button has low contrast (2.5:1). Increase contrast to meet WCAG 2.1 AA compliance.”
What’s amazing? These aren’t hallucinated hot takes, they’re grounded in design heuristics, accessibility guidelines, and real-world UI principles.
Get it directly on Figma with different plugins, like this:
Sure, human feedback still matters. Context, nuance, and taste, that’s where a good team excels. However, using AI for your initial draft of critique saves time.
A September 2024 study on arXiv showed that designers using LLM-based critique tools iteratively improved mockups, showing clear jumps in clarity, hierarchy, and usability over just three rounds.
Here’s a list of tools you can start using:
6. User flow optimization
I get it. There’s nothing more frustrating than pouring hours into a flow only to see users bounce halfway through.
Was it a clunky form? A confusing CTA? A missing next step?
Use AI to pinpoint friction, remove guesswork, and drive better UX decisions. Here’s what this looks like in real-time:
Take session replays, clickstream data, or analytics logs and feed them into tools like Hotjar, UXCam, or Mixpanel with GPT-based analysis. Then ask:
“Where are most users dropping off in this flow?”
“Can you map out what users do after they abandon this screen?”
“Suggest improvements to reduce friction in the conversion funnel.”
It will help you identify bottlenecks, optimize navigation, and craft smoother flows backed by data.
1. Can AI analyze user interactions to optimize navigation and flow?
Yes. Behavioral analytics tools powered by AI can automatically track how users move through your product and flag issues.
Tools that do this:
Hotjar + AI filters: Shows rage clicks, scroll drop-offs, and confusing navigation patterns.
FullStory: Uses machine learning to surface UX issues like “dead clicks” or friction events.
Mixpanel Predict: Identifies the most likely actions a user will (or won’t) take.
2. Are there AI tools that can predict drop-offs & recommend fixes?
Definitely. Predictive AI models can forecast where users are most likely to abandon a flow, and why.
Tools to try:
Contentsquare: Predicts drop-off points and suggests alternate flows or simplified UI.
Amplitude’s Pathfinder: Uses AI to map popular user journeys vs. dead ends.
Session AI: Personalizes content or flow in real time based on dropout predictions.
More prompt examples for designers using AI in user flow:
“Analyze this 4-step onboarding flow and suggest 3 changes to reduce drop-offs.”
“Where might users get stuck in this flow from homepage to checkout?”
“How can I reframe this form to reduce cognitive load?”
“Suggest improvements to my app’s menu hierarchy based on ecommerce best practices.”
Final takeaway
We’re no longer just pushing pixels or perfecting the kerning. We’re framing business problems, rewriting interfaces with tone in mind, optimizing user flows, and co-piloting with AI to build smarter, faster, and better.
But here’s the catch: the designers who win in this new era aren’t just “creative.” They’re curious, strategic, and relentlessly iterative. They don’t fear automation—they design with it.
So the real question isn’t “Will AI replace designers?”
It’s: Will you evolve fast enough to design what AI can’t?
Empathy. Context. Intuition. Storytelling. That’s still your edge.
Design isn’t just about what looks good—
It’s about what works. And works better every time.
That’s all for this week. See you with another interesting topic! 🙏
And if you enjoyed this breakdown, feel free to share & subscribe ❤️
ADPList’s Newsletter is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.
- Avani
Connect with me on LinkedIn