AI in Design: What Has Helped, What Hasn’t and What We’ve Learned

Amy Tolle
30 October 2025

Design has always been about experimentation: trying new approaches, noticing trends, and deciding which ones deserve attention. At Adage, our design team takes the same stance with AI. We don’t expect it to do our jobs for us, but we’re curious enough to test new tools, compare notes, and see where they can help us deliver more value. 

Along the way, we’ve learned two things:

  • Not every tool works for everyone. What sparks creativity for one designer might feel clunky to another.

  • The thinking and the end product still matter most. AI might clear away the tedium of layer naming or inspire a new direction, but the craft and strategy behind good design haven’t changed. 

Here’s a look at what we, the Adage Design Team, have tried, what’s worked, what hasn’t, and where AI is (and isn’t) making our work better. 

Allison Nance, Senior UX/UI Designer 

What's Helped My Work: NotebookLM

I like using NotebookLM because it helps me synthesize research and discovery results quickly. I can upload workshop notes, research findings, and interview transcripts all in one place, then ask about themes, specific questions, or user responses. It cites exactly where the information comes from, so I can always check the source. If the answer isn’t in my materials, it tells me instead of pulling from the web. 

I’ve used it to review hours of user testing for one client and to analyze discovery results that shaped a design strategy document. It saves time by turning large amounts of information into clear insights, and I appreciate that my data stays private. 

What Hasn't: OpenArt 

One tool that hasn’t worked as well is OpenArt, an image and video generation platform. I was drawn to its character feature, which promised consistency across images for storyboarding a client presentation. But I encountered too many inconsistencies. I tested Anime, realistic, and pen-drawing styles. While Anime performed best, the others had major issues features would change or styles would shift from one frame to the next. The realistic and pen styles I needed didn’t hold up. While I didn’t find it useful now, I’m hopeful it will improve.  

 
Jacob Perry, Senior UX/UI Designer 

What's Helped My Work: Figma AI Tools 

I’ve been skeptical about how AI could meaningfully assist me in my work, but one area where I’ve been pleasantly surprised is with the new AI features built directly into Figma. I’m not talking about Figma Make, but rather the practical tools woven into the everyday interface. Things like automatic layer renaming, generative content fill, and image resolution boosts have quickly become part of my workflow. 

They don’t replace design thinking or creativity; instead, they clear away the small, repetitive tasks that often slow us down. For me, that’s where AI really shines, not as a substitute for the craft and strategy of design, but as an assistant that removes friction, allowing me to stay focused on the bigger picture. With fewer roadblocks, I can reach a client-ready prototype faster and spend more of my energy on what matters most, the design itself. 

What Hasn't: ChatGPT 

ChatGPT hasn’t fully won me over just yet. It offers an impressive range of capabilities, including image generation, color palette development, and generative web design examples. These all have strong potential to support the early stages of refreshing a client’s visual design. So far though, I’ve found that most of these functions are handled more effectively by AI tools built specifically for those purposes. However, I’m continuing to explore and test ChatGPT to see how its features grow and improve over time.

Alejandro Londoño, Associate UX/UI Designer

What's Helped My Work: Adobe Firefly

Adobe Firefly is useful for generating images, photography, and digital art with prompts. It’s not quite on Midjourney’s level, but it works well if you know how to guide it. I like that it lets you choose styles, colors, and compositions up front, and you can upload reference images for closer results. I’ve used it for ideation, and quick visuals. I can trust the quality, and it saves me time

What Hasn’t: Gemini

Gemini, like ChatGPT, can be useful for a lot of things, but when it comes to image generation, I’ve found it a bit weak. Gemini can produce detailed, high-resolution images, but it struggles with applying specific styles, camera angles, or perspectives. It doesn’t keep consistent patterns, making it impractical for projects where visuals need to match. The results don’t look bad, but the lack of control limits its usefulness.

 

Anamaría Aguilar, UX/UI Designer

What's Helped My Work: Photoshop AI 

Photoshop’s AI tools make editing photos much faster. Manual tasks like cutting out objects, retouching, or adjusting backgrounds now take seconds. I use it to remove unwanted elements, adjust lighting, or generate creative variations. I value being able to experiment quickly and keep only what works, without losing image quality.

What Hasn’t: Figma Make

 Although promising, I haven’t found Figma Make to be useful yet. It doesn’t integrate smoothly with the rest of Figma, so the workflow feels isolated. Customization is limited, and I lose too much control over the final result. I often need external resources to reach the level of detail I want.

 

Natalie Callegari, UX/UI Designer

What's Helped My Work: Figma Make

Figma Make has become one of my favorite creative tools. With a decent prompt, it generates ideas that help me get past a blank artboard. I don’t use the designs as-is, but they spark directions I can riff on. I also use it with existing designs to explore alternate solutions. The AI provides a first draft, and then I do the fun part: experimenting and making it my own.

What Hasn't: ChatGPT 

ChatGPT is handy for quick copywriting or organizing scattered ideas, but it often misses details I’ve clearly provided, so I end up repeating myself. And its image generation always has that unmistakable AI “look.” For me, it works for text tasks if I’m attentive, but not for visuals.


Amy Tolle, Director of Design  

What’s Helped My Work: ChatGPT 

As a serial context switcher, I often have ideas quicker than I can sit and build upon them. ChatGPT makes it easy to build on ideas while keeping up with day-to-day work as well.  I often ask it to act in roles for me: act as a Senior Vice President for the children’s theater, or act as a copywriter specializing in UX writing to get more specific results. One of my favorite things to do once I’m satisfied with our results is to ask it to show how it thought about a problem that I posed – then, I can take it a little further on my own next time. 

What Hasn’t: Gemini 

I’ve used Gemini only for a niche task: book summarization. While it does the job, I find that I don’t retain the information as well, and often the insights turn pithy rather than applicable to my thought process. This is where the utility of AI becomes a burden for me; while it can accelerate knowledge transfers, ideation, and synthesis, if I can’t apply what I’ve learned later, the benefits are short-lived.

What We’ve Learned 

  • Experimentation is essential. AI is another arena for creative trial and error. Some tools hit, some miss, but testing them keeps us sharp. 

  • Tools work differently for everyone. What speeds up one designer’s process might not fit another’s. Diverse approaches make the team stronger. 

  • Our end goal is unchanged. AI doesn’t replace design thinking. It helps us get to thoughtful, richer, more tangible outcomes by broadening our scope, deepening our expertise, and expediting the work that took time, not thought.

Try It Yourself

If you’re not deep in design tools every day, here are broader lessons you can apply in your own teams: 

  • Stay curious (and willing to be wrong). The best discoveries come when you treat AI as an experiment, not a guaranteed solution. Build room for mistakes and surprises when exploring new tools. 

  • Focus on the prompt, not the output. Success often depends on how clearly you can frame the question. Practice writing better prompts by being specific, structured, and contextual. 

  • Think workflow, not replacement. AI shines when it’s part of a larger process: helping clear bottlenecks, generate options, or speed up research. It rarely replaces the human thinking, judgment, and creativity that make work valuable. 

At Adage, that’s the balance we aim for: curiosity without chasing the trend, experimentation without losing sight of outcomes, and a focus on where AI can add value as we create digital experiences that remove friction, encourage interactions, and delight users. 

AI is helping our design team move faster and think deeper, without replacing the craft behind great design. If you're exploring how AI could support your team’s workflow, we’d love to connect and share what we’ve learned.

Amy Tolle



Show more
Also of Interest:
Scroll to top