There’s a lot of buzz about AI agents. Robots that do more with less supervision—what could go wrong? We asked our community how this might shake up how we think about UX.

Welcome to the first installment of Double Click, a new series where we ask our community to weigh in on topics in tech and design that have bubbled up to the top of our feeds.

It’s official: The latest tech to rouse Big Feelings is agentic AI. You’ve probably seen the headlines about OpenAI’s Operator and Perplexity’s Assistant joining the party alongside offerings from Anthropic, Microsoft, and Google. Some corners of the internet are reacting with gusto. Olivia Moore, AI Apps Partner at Andreesen Horowitz, got Operator to pay a bill just by giving it a picture. “We are so back,” she posted on X. But others are experiencing “sheer unadulterated rage” at the idea of catering to trawlers, with one developer going so far as to build an infinite maze to trap them. So what does all this mean for how we create and consume sites? To get to the bottom of things, we asked our colleagues and collaborators to “double click” on the topic.

The great promise of agentic AI

Olivia Moore’s example captures why people are pumped about AI agents—namely, what they can take off our plates. Henry Modisett, Vice President of Design at Perplexity, points out that “AI should be how something works, not what you interact with.” Over Zoom, he brings up one of his favorite anecdotes about the R&D team at Apple: “They got a touchscreen working as a prototype and brought it to Jony Ive and Steve Jobs, who asked, ‘Okay, what do we do with it?’”

We’re in the what-do-we-do-with-it phase of agentic—and that’s energizing. “As a user, you don’t want the cognitive load of interfacing with a chatbot that has a name and a personality,” he continues. “You want to press a button and get the job done. Agentic reinvigorates the question of how much complexity is exposed to a user. It’s like starting a car—users don’t want to look under the hood to see how things are working. They want to just go somewhere.”

“Good design won’t just be about making things easier; it’ll be about reintroducing friction where it matters.”
_{Allan Yu, Co-founder, Synthetic Traffic}

Why this moment matters

All of this, of course, can feel a bit lofty. On a more practical level, Ben Hylak, Co-founder at Dawn Analytics, predicted on X that “optimizing/maintaining websites for agents will be an entire industry.” These agents won’t be impressed by fancy transitions or pixel-perfect layouts, says Founder and CEO of Bluecadet Josh Goldblum on LinkedIn: “Websites might evolve to favor clean APIs, machine-readable content, and streamlined navigation—less about aesthetics, more about function.”

Carly Ayres, Writer and Editor at Figma, raises the existential question: “Are we building for humans or machines? Sites were overdue for another shakeup. Despite the web’s roots in creative coding, it had gone the way of the templatized and the utterly depersonalized.

Maybe AI agents will push us to build better, more accessible websites. Or, perhaps we’ll all end up maintaining elaborate API documentation for our robot overlords.”

It seems we’re at a rare—and welcome—crossroads. Here’s the moment where we get to decide what work we want to sand away, and what we want to preserve. Konrad Burchardt, Growth Marketing Manager at Figma, points out that making nice with AI shouldn’t be the only priority: “What happens when we use AI to supercharge everything?