Figma Config 2026 just wrapped. Our founder and CEO Tejas Shah spent three days in San Francisco at the biggest design conference of the year. One word kept showing up more than AI did. Agents. And what that actually means for how design teams work. 18 years of design engineering at Enspirit, long before AI made the path shorter. #Config2026 #Figma #ProductDesign #AIAgents #DesignEngineering #Enspirit
Figma's biggest conference of the year just wrapped. Something I noticed in the keynotes: they kept saying "Agents" instead of "AI." Not once or twice. Consistently, across the whole program. It's a small choice doing real work. AI is a tool you buy. An agent is something you direct. It produces, and someone has to read what comes back and decide whether it's any good. That someone needs judgment. The word quietly puts a human back at the center. A design studio founder made the case I've been making for years: design needs to own the code. He described how AI compressed what used to be a multi-year jump from designer to design engineer into about four months. Fair enough. But the idea underneath it, that the real work is closing the gap between what a designer intends and what actually ships, is not new. We have worked that way since 2008. Mobile apps when smartphones were new. Touchscreen TV interfaces. Kiosks, digital signage, product configurators. PLMs, ERPs, enterprise travel platforms. On one of those travel platforms, we took a booking flow from 425 seconds down to 127. Not by redrawing screens. By owning what reached the user. All of it before AI made the path shorter. We hired for design engineering when almost nobody was, trained for it with no map, and built the studio on one belief: the space between a designer's intent and the shipped product is where quality usually dies. AI didn't hand us that belief. It just made it harder to ignore. The teams that do well next won't be the ones generating the most. They'll be the ones who can look at what an agent produced and catch what's wrong with it. An agent has no taste and no stakes. It can't tell that a flow which passes every test will still frustrate a real user. A person catches that, and only if they've seen it break before. There's a harder thing an agent can't do at all. It can't feel where users are going. It works from what already exists, so it's always pointed at the present. A person who has stood in the user's shoes senses the shift before the data shows it, the rising expectation, the thing that felt fine last year and won't next year. That instinct is where good products come from. We have eighteen years of seeing things break, and of watching what people come to expect next. #Config2026 #Figma #ProductDesign #AIAgents #DesignEngineering #Enspirit #UXDesign #DesignSystems #B2BSaaS Enspirit