What Adobe Knows About AI That Most Tech Companies Don’t

Last week, I was talking with a graphic designer about Adobe MAX, and they shared with me the most unexpected review of an AI feature I’ve ever heard. “Photoshop will rename your layers for you!” he said, without hesitating. The feature he was referring to was that Photoshop can now look at the content on each of your layers and rename them for you. Since most people don’t give a lot of thought to naming layers as they create them, this might be one of the most useful features Adobe has ever created. It’s certainly one of the most useful AI features that any company has come up with so far, mostly because it does something very helpful but that no one wants to do. Helpful over hype And, that’s the point. In fact, that reaction explains more about Adobe’s AI strategy than anything the company demoed during its keynote. It’s not the kind of feature that gets a lot of hype, but I don’t know anyone who regularly uses Photoshop who wouldn’t prefer to have AI handle one of the most universally hated chores in design: cleaning up a pile of unnamed layers. I think you can make the case that Adobe just made the loudest, clearest argument yet that AI isn’t a side feature. In many ways, it is the product now. Almost every announcement touched Firefly, assistants that operate the apps for you, “bring your own model” integrations, or Firefly Foundry—the infrastructure layer that lets enterprises build their own private models. What Adobe understands But beneath it all, Adobe is doing something most tech companies still aren’t. Instead of looking for ways to bolt AI onto its products, Adobe is building AI into the jobs customers already hired Adobe to help them do. When I sat down with Eric Snowden, Adobe’s SVP of Design, at WebSummit this past week, he used a phrase that stuck with me: “utilitarian AI.” Sure, there were plenty of shiny new AI features that Adobe announced like Firefly Image Model 5, AI music and speech generation, podcast editing features in Audition, and even partner models like Google’s Gemini and Topaz’s super-resolution built directly into the UI. But Snowden lit up talking about auto-culling in Lightroom. “You’re a wedding photographer. You shoot 1,000 photos; you have to get to the 10 you want to edit. I don’t think there’s anybody who loves that process,” he told me. Auto-culling uses AI to identify misfires, blinks, bad exposures, and the frames you might actually want. Ultilitarian AI is underrated That’s what he means by utilitarian AI—AI that makes the stuff you already have to do dramatically less painful. They force you into an “AI mode,” but instead save you time while you go about the tasks you already do. Snowden describes Photoshop’s assistant like a self-driving car: you can tell it where to go, but you can grab the wheel at any time—and the entire stack of non-destructive layers is still there. You’re not outsourcing your creative judgment—you’re outsourcing the tedious tasks so. you can work on the creative process.. That’s Adobe’s first insight–that AI should improve the actual job, not invent a new one. The second insight came out of a conversation we had about who AI helps most. I told Snowden I have a theory: AI is most useful right now to people who either already know how to do a thing, or don’t know how to use the steps but know what the result should be. For both of those people AI helps save them meaningful time. That’s how I use ChatGPT for research. I could do 30 Google searches for something, but ChatGPT will just do them all at the same time and give me a summary of the results. I know what the results should be, and I’m able to evaluate whether they are accurate. The same is true for people using Lightroom, Photoshop, or Premiere. You know what “right” looks like, so you know whether the tool got you closer or not. AI can do many of the tasks, but it’s still up to humans to have taste. AI has no taste Which is why Snowden didn’t hesitate: designers and creative pros are actually better positioned in an AI world—not worse. “You need to know what good looks like,” he told me. “You need to know what done looks like. You need to know why you’re making something.” Put the same AI tool in front of an engineer and a designer and, according to Snowden, “90 times out of 100, you can guess which is which,” even if both are typing prompts into the same tool. That means taste becomes the differentiator. Snowden told me he spent years as a professional retoucher. “I think about the hours I spent retouching photos, and I’m like, I would have liked to go outside,” he said. Being able to do that skill was important, but it wasn’t the work. The finished product was the work, and AI can compress everything between the idea and the result. Trust has never mattered more The third thing Adobe understands—and frankly, most companies haven’t even started wrestling with—is trust. I have, many times, said that trust is your most valuable asset. If you’re Adobe, you’ve built up that trust over decades with all kinds of creative professionals. There is a lot riding on whether these AI tools are useful or harmful to creatives, as well as to their audiences. So, Adobe didn’t just ship AI features; it is building guardrails around them. For example, the Content Authenticity Initiative will tag AI-edited or AI-generated content with verifiable metadata. Snowden’s framing is simple: “We’re not saying whether you should consume it or not. We just think you deserve to know how it was made so you can make an informed choice.” Then there’s the part most people never see—the structure that lets a company Adobe’s size move this fast. Understanding how customers want to use AI Snowden’s team actually uses the products they design. He edits photos in Lightroom outside of work. Adobe runs a sort of internal incubator where anyone can pitch new product ideas directly to a board. Two of the most important new tools—Firefly Boards and Project Graph—came out of that program. When AI arrived, Adobe already had the mechanism to act on it. It didn’t need to reinvent itself or reorganize. It just needed to point an existing innovation engine at a new set of problems. That’s the lesson here: Adobe isn’t chasing AI because it’s suddenly trendy with features no one is sure how anyone will use. It saw AI as a powerful way to improve the jobs its customers already do. That’s the thing so many tech companies still miss. AI is not a strategy. It’s not even the product. It’s a utility—one that works only if you know what your customers are trying to accomplish in the first place. So far, it seems like Adobe does. And that’s why its AI push feels less like a pivot and more like a product finally catching up to the way creative work actually happens. EXPERT OPINION BY JASON ATEN, TECH COLUMNIST @JASONATEN