How to harness skills that AI can’t automate

In a world where anyone can prompt their way to a working prototype, what sets good products apart from great ones is the care and quality that goes into them. Without thoughtful human input, AI tools generate outputs that work on paper but miss the mark in practice. Bridging that gap requires distinctly human capabilities like curiosity, intuition, taste, and intention. Here, we dive into how you can cultivate and leverage these skills so you can steer AI tools toward quality outcomes, not just quick ones.

Curiosity: Starting from first principles

Curiosity is asking “why” and “what if” before a clear path has been carved out; it’s considering dozens of different directions, then iterating until you validate the right one. Traditionally, this kind of curiosity takes time, but AI tools have lowered the barrier to exploring more directions. With Figma Make, for example, a product manager can enter a prompt, generate prototypes, and see which ideas are worth pursuing in earnest—all without writing a line of code. A marketer using Claude or ChatGPT can spin up several campaign outlines in minutes, comparing them to see which one aligns most closely with their brand voice and objectives. But even with these accelerants, there are still real limitations. AI can’t decide to wander away from the original scope just to see what might happen, or question whether the scope itself should change. It can’t feel the pull of an unexpected idea or a hunch, or sense when something deserves deeper exploration, even when it’s time-consuming.

Curiosity doesn’t just expand the number of ideas on the table—it sharpens the confidence in the ones you choose. Exploring broadly at the start makes it easier to refine later, with fewer second guesses and rework. Take, for example, Figma Product Designer Natasha Tenggoro, who used Figma Make to test and validate different ways to add videos to Figma Buzz. Spinning up prototypes, she says, “Let me explore more directions, test edge cases early, and refine with confidence.” This helped engineers understand the feature’s scope and feasibility, too, avoiding costly back-and-forth for the whole team.

Learn more about how Product Designer Natasha Tenggoro used Figma Make to prototype early experiments for adding video playback to Figma Buzz.

How curiosity reshaped Figma’s icon suite

As Figma prepared to roll out four new products—Figma Make, Figma Buzz, Figma Sites, and Figma Draw—product designer Tim Van Damme faced the task of creating icons that not only represented each tool clearly, but also worked together as part of a consistent visual language.

On the surface, the project seemed straightforward: Design four new icons to represent four new products. But as Tim began sketching out potential directions for each of them, he realized he needed to take a step back. What if, instead of focusing on each icon individually, he redesigned the entire suite of Figma icons for a more updated, unified feel? It would add weeks to the project timeline, but a more holistic reimagining felt worth it. He expanded the scope into a full overhaul, revisiting existing icons alongside the new ones, exploring hundreds of variations, and making sure every symbol felt distinct on its own—yet cohesive as part of the broader Figma icon family.

The refreshed icon suite is both more unified and more distinctive. Each icon clearly represents each product, scales seamlessly to large formats like billboards, and remains unmistakably Figma.

Learn more about Figma Product Designer Tim Van Damme’s approach to icon design and see the explorations that shaped final results.

Intuition: Following a feeling

While AI can speed up time to market, it can’t tell you what will resonate best with a customer—but product builders can. They can sense a confusing interaction, even if it technically works in the prototype. That same instinct might lead them to advocate for more white space in a layout that feels cramped, even when all the content fits within the given constraints. “True design is when you’re facilitating a vision of what could be,” says Eliel Johnson, Vice President of User Experiences and Design at CVS Health. “I like to think about design as a verb, not just a job title. We want to be designing and asking, ‘Does it feel good?’”

What sets products apart today isn’t just their functionality, but their ability to spark an emotional response. Achieving that level of emotional resonance requires anticipating what will truly land. “Attempts at building moats through features won’t work anymore,” writes Figma’s Head of Insights, Andrew Hogan. “If your team can’t intentionally design for an emotional response, competitors will copy your functionality at a fraction of the cost. This means understanding users’ deep fears, hopes, and motivations that are difficult to get at—and often only visible outside of a screen.” In other words, designing products that connect on a deeper level requires emotional intelligence and instinct.

How intuition guided Plaid’s rebrand

After years of helping users securely link the apps they use to their bank accounts, Plaid expanded their product offering to identity verification, financial insights, and fraud prevention. These new products widened their audience beyond consumers to include banks and regulators—each with distinct needs and challenges. This evolution required a visual approach that could nod to this product expansion and speak to different audiences while still feeling cohesive.

Plaid initially partnered with an external agency on the rebrand, but capturing the nuances of its audiences and how the brand might evolve required a level of context only the internal team could bring. The project soon came back in-house. The team had already lost months, and working solely with internal designers and marketers would suck up even more of the team’s bandwidth. Still, they believed that putting the work in the hands of internal experts who understood their audiences and product deeply would pay off.

The team thought anew about visual directions, testing visual patterns drawn from currency and experimenting with holographic effects. They shifted to a faster, more collaborative rhythm, but the new approach came with tradeoffs: Moving faster meant embracing gut feelings instead of lengthy research cycles, and exploring broadly meant risking more time burned. “It’s like building a Lego set and saying, ‘I’m going to build a house,’ without knowing exactly what the final house will look like,” says Christophe Tauziet, Head of Design at Plaid. “You just keep going and make changes along the way.”

In the end, the tradeoffs were worth it, resulting in a brand identity flexible enough to grow alongside Plaid—one that’s modern and distinctive, yet grounded enough to build trust with consumers, banks, and regulators.

Taste: Sweating the details

Where intuition is a gut feeling that something might work, taste is the ability to judge which ideas to refine and which to discard. “Most companies confuse taste with aesthetics,” writes early Figma investor Sarah Guo. “But real taste runs deeper—it’s in the error messages, the loading states, the features you killed because they were merely good, not essential.”

Taste is killing your darlings, like Tim did when he scrapped the idea of a bee icon for Figma Buzz entirely, even though he’d spent days designing hundreds of iterations. Taste is knowing what to prioritize when time is limited, like the Duolingo Math team did by polishing core interactions instead of adding extra features when testing new math games. It’s knowing when to stop tinkering and ship it—even if it feels like there are endless refinements you could make. As Sarah puts it, “Real taste hurts. If your ‘taste’ doesn’t cost you something, it’s not taste. It’s preference.”

As AI expands the number of ideas a team can generate, taste becomes the editor, the sense of discernment that’s innately human. AI can follow rules, enforce consistency, and even spot visual irregularities, but it can’t weigh which of two technically correct solutions creates the most trust or delight.

Refining horizontal scrolling in Figma’s Layers panel

Designing the horizontal scroll bar in Figma’s Layers panel—the option to move sideways to reveal layers that don’t fit in view—came with myriad small decisions and considerations. The panel needed to handle thousands of elements with shifting states without disorienting users or slowing down their work. The challenge was finding a solution that felt both functional and effortless.

Figma’s design and engineering teams built quick prototypes to see how different ideas held up as users scrolled through thousands of layers with shifting states. Some options looked promising on paper but fell apart in practice—for instance, automatically jumping to a layer made people lose their place in their work, and adding small markers to show hidden layers only cluttered the panel. Other questions, like how much blank space to leave above visible layers, didn’t have one “right” answer. To make those decisions, the design and engineering teams iterated and refined until the interface felt balanced and unobtrusive.

The final design balances subtlety and clarity with thoughtful white space and heuristics that determine when—and how far—to auto-scroll when users click deeper into layers on the canvas. The result feels seamless to users, but was shaped by meticulous judgment applied to the smallest details.

Adding a horizontal scroll bar to the Layers panel in Figma proved a complex undertaking. Read more about the explorations that led to the final result and the lessons along the way.

Intention: Designing systems for scale

While AI tools can accelerate design work, they’re only as good as the context you provide. Design systems package craft into a consistent framework that AI can understand, reducing rework and preventing quality or design fidelity issues. For example, AI can’t guess that your team prefers 8px padding instead of 12px, or that you never use red-500 for CTAs. AI also can’t tell when a brand’s growth calls for its design system to be reimagined with new components, colors, or typography. It takes design expertise to create a design system that will give AI the guardrails it needs to generate usable outputs.

As Marcel Weekes, Figma's Vice President of Product Engineering puts it: “[AI] is really good at understanding the structure of a design system. You want your website or your application to look like your identity, so being able to provide that context of your design system to AI is worth investing in.”

How design systems helped shape Polaroid’s new era

As Polaroid reinvented itself for the digital age, its design work was slowed by fragmented tools and inconsistent workflows. Designers worked in isolated files, and there was no centralized design system to unify the brand across products. The friction was so disruptive that the team only designed an IOS app.

To solve this pain point, Polaroid’s UX team began building a design system in Figma to serve as a single source of truth. They started by building shared components, tokens, and variables to manage colors, fonts, and themes across platforms. From there, they standardized core components so iOS and Android designs could be maintained side by side, cutting down on duplicate work. The result is the system the team relies on today. With it, Polaroid now ships faster and with greater consistency. Designers can scale work across multiple products and platforms, and developers work from the same set of rules, ensuring outputs remain cohesive and unmistakably Polaroid.

Learn how Polaroid is reaching a new generation of users with a fresh approach to their design systems and workflows.

In the age of AI, speed and scale are easy to come by. What’s harder, and far more valuable, are the human choices that make something feel artful and alive. As Sarah puts it, “In a world where AI can instantly generate a CRUD app or replicate any website, taste becomes the final differentiator. Features can be copied. Functionality can be matched. But the feeling of using something crafted with intention? That's irreplaceable.”