Design Leaders: Speed Isn’t Enough Without Human Judgment.

Inside the growing effort to tether generative artificial intelligence to rigorous design systems without abandoning human perspective.

The growing promise of generative artificial intelligence is simple: create digital products faster, from early sketches to working code. Tools and platforms now advertise one-click screen creation and rapid assembly of user journeys. Yet in the real world of product development, many teams are discovering a sharper truth. AI can move quickly, but it does not always deliver what users truly need.

This gap between speed and usefulness is becoming clearer to designers, engineers, and product leaders who are working under real constraints, including tight deadlines and real human stakes.

Silu Manandhar’s Design Challenge in Digital Health: Using AI as a “Second Brain” to Explore a new medication-support feature Without Losing Care or Accuracy

Silu Manandhar, who designs digital health systems for Lose It! at the Everyday Health Group, encountered this challenge while developing a sensitive new medication-support feature. The team had to map many interconnected user situations across both iOS and Android. That work included onboarding to a new feature, medication dosage logging, customization of the feature visibility in the app, and personalized nutrient dashboards.

The timeline was aggressive. The team needed a way to explore many user paths without losing accuracy or care.

Rather than treating generative AI as a fully autonomous designer or dismissing it as a novelty, Manandhar built it into her process. She calls this approach her “second brain.” She used large language models to explore different user perspectives, test edge-case scenarios, and produce early low-fidelity layout ideas. With this support, the team cut an exploratory wireframing phase that could have normally taken 14 days down to 5 days. Those saved days were not used to ship a rushed or unfinished feature. Instead, the time was redirected to the kinds of decisions automated tools often get wrong. That includes handling sensitive user emotions related to weight, balancing business goals with accessibility needs, and pressure-testing how the product works with engineers before it reaches users.

For Silu, this is where the real nature of design shows itself. Design, she argues, is never just about how something looks. It means understanding human psychology, how people behave, and how they interact with digital products, which keeps shifting as technology opens up new ways to interact. Every screen is a web of decisions, balancing what is good for the business with what is good for the user, weighing technical constraints against opportunities, and judging how it should all look and feel. That critical thinking, she says, is the part of design people rarely talk about. AI made the process faster, but faster is not the same as right. The most important decisions, what to build, how an interaction should feel, how to respect a user’s emotions and sensitivities, still come from a person. There is an intangible sense of how someone will react that AI simply cannot catch. AI is a second brain, she says, not the main one.

Why automation struggles with the details that matter

The project also showed where AI falls short in product design. Many generative tools work by spotting common patterns from large amounts of existing content online. That can help when you are building something familiar, such as a standard settings screen.  But real products often depend on details that do not follow a simple template. User journeys are shaped by the situation a person is in, the device they are using, and the emotions they may feel while using the app.

When AI suggests an interface for those more complex situations, the result can look polished while still missing the most important pieces. It may include elements that do not belong, overlook platform-specific expectations, or fail to reflect the emotional reality of the users.

AI also cannot truly understand why a design choice feels safe to one person and upsetting to another, or how small changes in text and spacing can affect readability for older users.

Building guardrails to keep ideas usable

To prevent AI from drifting into ideas that did not fit the real product, Manandhar added engineering guardrails to her workflow and complemented them with clear documentation. She connected her company’s standardized iOS design system directly into Claude, providing access to the team’s component libraries, spacing rules, color guidelines, and platform specifications, so AI-generated concepts stayed within valid, production-ready building blocks. 

AI could still help the team explore quickly, but the exploration remained grounded in the structure of the actual app. She also maintained a structured design decision log in Figma, where designers record the reasoning behind each choice, the relevant research, and the trade-offs involved. When software assets can be generated in seconds, documenting why something was chosen becomes even more important, because faster creation produces more options and clearer decisions help teams select the right direction, not just the quickest one.

Manandhar shared the workflow beyond her immediate team. She presented it at company technical forums, featured in the company’s internal AI newsletter, and poke on a panel for Iowa State University’s Human-Computer Interaction program on how user experience evolves as AI tools become more capable.

The takeaway for product teams

Design leaders are aligning on one point: faster does not mean better. Most apps compete in crowded markets, and features often look alike, but the products that endure are those built with a consistent grasp of user needs, real-world context, and clear intent.

AI can speed up early ideas, but people are still accountable for the final result. The details, from interface choices to wording and interaction behavior, require human judgment.

About the Author

Sashindra Suresh is an experienced writer specializing in artificial intelligence, software development, and emerging technologies. With a strong ability to translate complex technical concepts into clear, engaging insights, she has contributed to a wide range of publications and platforms. Her work focuses on making cutting-edge innovations accessible to both industry professionals and curious readers alike.