Over the past couple of years, I’ve been gradually weaving AI tools into my day-to-day work as a designer. With the same number of hours in a day, it now feels like more can be done, but in a sustainable fashion. These tools help me be more creative, communicate more clearly, make sense of complex information, and show an idea in its realest form.
I use AI in three main areas.
Tools like GPT have been especially useful. Beyond the obvious tasks like cleaning up writing, or drafting documents, I’ve found deeper value in how they help with structuring messy information, identifying patterns in feedback, and reaching a clearer starting point more quickly.
I’ve shared a few examples of how I use AI regularly below. Each one plays a different role, depending on the kind of input I’m working with or the outcome I’m aiming for. I am continually learning how to work more effectively with AI, as the technology is constantly evolving. And it's not automatic, AI requires judgment, oversight and intentionality.
Turn technical docs into design briefs.
Tool used: ChatGPT
PM specifications and engineering documents often arrive in a state that’s either too technical or too ambiguous to work from. I used to spend a lot of time reworking them, reformatting, rewriting, and trying to clarify what actually needed to be designed. I started using GPT to help rephrase and restructure these documents into something more closely resembling a design brief. While it still requires judgment and editing, it saves time and helps me reach a clearer, more usable starting point.
Making sense of feedback and doing competitive analysis.
Tools used: ChatGPT, Claude, Notebook LLM
I sometimes work with raw feedback, including interview transcripts, customer comments from help forums, and public reviews from sites like G2 and Reddit. GPT and Claude have helped make this more manageable. I often copy and paste blocks of feedback or user reviews into GPT, which enables me to identify recurring themes, extract illustrative quotes, and organize the information into a digestible format. Claude organizes content as “artifacts,” and GPT enables me to create custom workspaces that preserve context and facilitate ongoing exploration of the data.
However, Notebook LLM has become my go-to tool for accuracy. It reduce hallucinations and cites the source of the information.
Turn information into something you can listen to and learn from
Tools used: Notebook LLM (Google)
Notebook LLM has become one of the most useful tools in my process when I’m dealing with dense or complex material, especially research transcripts. You can also collect a batch of research papers or competitive data and use Notebook to explore patterns, surface insights, or make the material more approachable and easier to work with. It’s more reliable than tools like GPT for tasks that require accuracy or traceability, as it cites the source of each insight and is less prone to hallucination.
After uploading content, I often use the “Audio overview” feature to generate a two-person podcast-style discussion about the material. It’s a great way to passively absorb information and hear it reflected back in a conversational tone. You can also ask questions during playback, and the AI podcast hosts will respond in real-time.
Notebook also offers other formats that can be helpful, depending on the situation, such as study guides, briefing documents, FAQs, and timelines. These provide different ways to break down or structure information, especially when trying to distill something complex or prepare it for others.
Simple design feedback
Tool used: ChatGPT
Working as an IC, even at a large company, can be isolating. Designers need feedback, but without regular critique sessions or dedicated time from teammates, it’s not always easy to get. I’ve used GPT to help fill that gap, especially when working on more technical projects where I want to pressure-test my thinking or look for oversights.
While I don’t rely on it exclusively (and I don’t expect it to give the best high-quality feedback like a designer might), I’ve been surprised by some of the things it catches. It’s especially useful for pointing out accessibility issues or raising questions about clarity and flow, areas where a second set of eyes can make a big difference. It’s not a substitute for human feedback, but it’s a solid first pass.
I started using Midjourney in 2023. At first, I was just experimenting, but I quickly saw there was a learning curve, so I enrolled in Midjourney for Creatives, a course by Nick St. Pierre. Since then, I’ve been casually exploring and following other creators. I’m constantly in awe of what’s possible and what people are making with it. It also made me realize how much I didn’t know. Understanding art history, various genres, and referencing specific artists can significantly enhance the results. When you follow great Midjourney creators, you start to notice how often they draw from established styles and artists to guide their prompts. While I haven’t used Midjourney in my product design work yet, I’ve used it to create nearly all of the images on everyday ux.
Most recently, I’ve been using AI tools like Vercel, Claude, and Figma Make to prototype and build working software. Before these tools were available, I relied on Framer to bring ideas to life. Now, the gap between concept, design, and execution feels smaller than ever. These AI tools have been a game-changer, especially because, as most designers know, the more tangible an idea is, the greater its impact can be. This is just another step toward making ideas real, faster.
Messenger widget customization app
Tool used: Vercel
While at Zendesk, I built an internal tool for the Voice of the Customer (VoC) team to customize the Zendesk Messaging Widget. Personalizing the widget is a key part of how the team works with prospective customers; it helps them showcase what the experience could look like in context. Before this tool existed, the process was manual, time-consuming, and frustrating. The new tool replaced that workflow entirely, making it faster and easier to tailor demos.
Admin center homepage
Tools used: Figma Make
I used Figma Make to bring a vision concept to life—starting with static design mocks and quickly turning them into a working prototype. Once the first version was in place, it was easy to layer in new components and widgets. Through prompting and iteration, I refined the design and arrived at a realistic, high-fidelity version of the idea.
DJ Mixes app
Tools used: Vercel
I’ve also been using these tools to follow my curiosity and build small apps for fun. One of them is a DJ mixes app, inspired by a corner of YouTube where people DJ from their living rooms, cafés, bakeries, or beautiful outdoor spaces. These mixes had become my personal bookmarks, and I wanted a simple way to bring them into my living room.
3D image prompt library
Tool used: Vercel
ChatGPT has become surprisingly good at image creation, and a growing number of creators are developing their own styles and prompt techniques. I’ve been bookmarking my favorites and built a small gallery to collect them. Each entry includes sample images, the original prompt, and often JSON instructions. You can browse the gallery, explore different styles, and easily copy prompts to create your own.
Need advice on how to use AI in your company?
Feel free to get in touch anytime, I'm happy to help.