How to design AI products that feel human
What does it take to design an AI product people actually want to use every day? For Henry Modisett, founding designer at Perplexity, the answer isn't in flashy features or cutting-edge tech. It’s in clarity, speed, and deep respect for the user’s time.
Start with systems, not screens
When Modisett joined Perplexity, it was barely more than a tech demo. His first move wasn’t to design mockups, it was to build a flexible component system in React. Without a defined product, he focused on enabling fast iteration through reusable UI primitives like buttons, grids, and typography. That investment let the team pivot quickly and ship prototypes that went viral, even without a launch strategy.
Designers who code = superpower
At Perplexity, designers are expected to write production code. Modisett believes this cuts through red tape, enables polish, and fosters a deep emotional connection to the product. “If I see a bug, I fix it,” he says. “It’s not a process problem, it’s an empowerment problem.”
Culture > process
Rather than rigid critiques or top-down design mandates, Modisett builds teams with founding-designer energy: gritty, versatile generalists with conviction. He avoids over-exploration, debate, and visual nitpicking in favor of fast decision-making and constant iteration. “Most feedback halfway through isn’t that helpful,” he says. “Just get it working and obsess from there.”
Consumer-first, AI-last
While most AI tools emphasize tech, Perplexity focuses on value. The team deliberately avoids talking about AI in the product. “Users just want answers,” Modisett explains. “They don’t care how it works, only that it works, and fast.”
Brand as a recruiting tool
Perplexity’s brand isn’t just for users, it’s for attracting great talent. By investing early in distinct visual identity and working with non-tech brand agencies, the company stood out while still feeling serious. The brand is deliberately flexible, designed to reflect the personalities of the designers building it.
Designing generative interfaces
Looking ahead, Modisett envisions Perplexity as a platform that dynamically assembles interfaces based on user intent. Whether it’s comparing stock prices or planning a trip, the UI adapts in real-time to present the best possible format, charts, forms, or summaries.
A new era for designers
Modisett sees the rise of AI as an opportunity for designers, not a threat. With foundational AI capabilities now commodified, design becomes the differentiator, just like it once was for radios. “We’re heading into an era where form and function in hardware and software will matter again,” he says. “Design is becoming a competitive edge.”
Listen to the full episode here.
Source: Dive Club podcast
You might also like
