A Q&A With the Co-Author of Our Latest Report on AI Design Tools

With the rise of generative AI tools, design leaders are feeling the pressure to move fast, but often without a clear strategy or path. We wanted to see how these tools actually perform in real design workflows.

Principal Designer Scott Gerlach sat down to talk about the recent report he co-authored, which evaluates the current generation of Gen AI tools through the lens of real product design work. Below, he shares what surprised him most, what the tools get wrong, and what design teams need to keep in mind as they experiment.

Hey Scott, tell me about the report you worked on.

Sure. I'm Scott Gerlach, a Principal Designer here at argodesign, and I recently worked on a project where we evaluated some of the latest generative AI tools and how they relate to the design process. I then co-authored a report with my colleague Thomas Sonsalla to share our findings.

Why is this report so important, especially right now?

From my perspective, the report is significant because it doesn’t chase the hype about what Gen AI might be able to do in the future. Instead, it offers a clear-eyed look at what happens when we try to use current Gen AI tools within a real design process—when we're actually trying to produce a product, not just isolated tasks.

What surprised you the most when researching this report? Did you come out of it thinking something different than when you went in?

You know, it's interesting. If I’m being honest I was surprised at how ineffective most of the tools are when it comes to supporting designers in tasks related to creating meaningful UI that supports real workflows.

There’s a common mental shortcut we all take when thinking about what Gen AI should be capable of. We think: there are large language models that are good at language, and there are large models that are good at image generation—so isn’t creating UI just those two things combined? But the reality is, what you need is a language model that truly understands interaction, and that’s a much more complex concept.

You quickly run into the limitations of current tools trying to generate UI. What you get is often either images pretending to be UI, or code and language pretending to be UI—but very little that actually bridges the two in a functional way.

Is this something you and other argo designers have been talking about a lot—on projects, in one-on-ones, at the lunch table? Does it come up when talking to others in the industry?

I think what we encounter a lot as designers—especially at a design agency—is two things. First, we’re often working with clients who are feeling pressure from within their organizations to be more efficient by leveraging AI. Much of what we do involves investigating whether and how that might be possible.

So a lot of our conversations are about specific things we can do to help a client. But as designers, many of us carry these nagging doubts about the broader hype around Gen AI and what it actually means for our day-to-day work in product design. And I don’t often hear people talk about that part in much detail—in conversations, in articles, or anywhere that breaks it down step by step.

What value does this report offer design leaders or executives who are feeling pressure to implement Gen AI without a strategy or plan?

I think the value of the report for design leaders is twofold.

First, it gives them ammunition—to say, “Here’s a case where someone has taken a close look at what’s possible and practical.” That lets them engage more responsibly with other parts of their organization when discussing what's realistic in terms of improving efficiency through design right now.

Second, it grounds them in reality. It helps them think through two sides of the equation: What’s the cost of investing in tools that aren’t yet ready for real workflows? And how do you prepare your team for the near-term—so they’re efficient today, but also have the right mindset for how Gen AI will become part of their workflows over time?

Why is "disruption" the wrong lens for design teams to apply to Gen AI?

There’s something interesting happening with Gen AI that runs counter to our experience in design. Take Figma, for example. That was a radical shift from previous tools. But it wasn’t disruptive in the sense of breaking everything. It was adopted because it was practical—because it solved real problems and enabled teams in ways existing tools didn’t.

The same is true for any tool. In enterprise and team settings, we should focus on tools that can be integrated into a scaled, repeatable practice—not those that promise to upend everything overnight. In reality, things only get adopted when they fit into the day-to-day—not when they demand that we throw the day-to-day out because someone said a new tool is important.

What’s the right mindset for teams experimenting with Gen AI today?

If your team wants to experiment with some of these tools, the first thing is: don’t go it alone. If leadership just says, “Hey, these new tools exist—go try them out if you want,” what that really means is, “Make extra time in your day to learn something new and be efficient with it.” And that’s not a realistic ask.

A better approach is to define a shared format: How will we look for opportunities to explore a tool? How do we account for the extra time and effort that exploration takes? And how do we share what we learn in a structured way?

This helps the whole team know what’s expected—and avoids leaving people feeling like they’re falling behind or unsure whether they should be doing something on their own. Without that shared structure, you end up with people experimenting in isolation and no real way to integrate what’s learned back into the team’s workflow.