When the Plugin Is a Prompt, Knowing Figma Is No Longer Enough
As AI makes tool fluency cheap, judgment becomes the skill that still matters.For about fifteen years, “knows Figma” was a line you could put on a résumé and have it mean something. It meant you could move fast in the tool everyone else used, that you spoke the shared language of components and…
As AI makes tool fluency cheap, judgment becomes the skill that still matters.For about fifteen years, “knows Figma” was a line you could put on a résumé and have it mean something. It meant you could move fast in the tool everyone else used, that you spoke the shared language of components and auto-layout, that a hiring manager could hand you a file and trust you not to break it.At Config 2026 on June 24, Figma spent a keynote quietly draining the meaning out of that line, and then on July 7 it bought a company that finished the job. I have been turning the two events over for a couple of weeks now, and I keep landing on the same uncomfortable conclusion: the thing a lot of designers were actually being paid for was fluency in a piece of software, and that fluency is about to be worth roughly what fluency in a specific keyboard shortcut is worth now.Here is what Figma actually announced, because the specifics matter more than the vibe. Code Layers let you turn a design layer into interactive code in one click, with the coded prototype running side by side with the static one, rolling out from July via waitlist. Generative Plugins let you describe a tool you want in plain language and have the Figma Agent build it on the spot.There is a keyframe motion timeline that exports to real formats, node-based workflow canvases, shader fills generated by AI, and an agent wired into Notion, Slack, and GitHub with web search. The keynote tagline was “no tool should limit where an idea can go.” The design file, in other words, stopped being a document you hand to an engineer and started being a runtime that produces the software itself.The two moats that just cracked in the same monthThere were always two separate moats hiding inside the phrase “knows Figma,” and it helps to pull them apart, because they broke for related but different reasons.The first moat belonged to the designer. Tool proficiency was a real, hireable, defensible skill, because the tool was hard and mastery took years. You could build a career on being the person who could make Figma do things other people could not, the same way an earlier generation built careers on knowing the darkroom or the typesetting rig.When Generative Plugins mean the specialized workflow you spent two years mastering can be described in a sentence and generated in a minute, that specific mastery stops being scarce. The person who “knows Figma” deeply and the person who can describe what they want clearly are converging on the same output, and only one of those two skills took years to acquire.The second moat belonged to Figma. For a long time the company’s defensibility was that everyone already knew it, that switching cost was enormous because the whole industry had standardized on its file format and its muscle memory. That moat is exactly what erodes when plugin development becomes a prompt, because the specialness of the tool was partly the accumulated cleverness of its plugin community and its learned complexity, and both of those are the kind of thing an agent flattens.When any tool can generate any tool, being the tool everyone knows is worth less, because knowing the tool is worth less. Figma appears to understand this better than its users do, which is why the July 7 acquisition is the more revealing of the two moves.Buying the thing that eats youTwo weeks after Config, Figma acquired the team behind Bud, formerly Orchids, a Y Combinator vibe-coding startup run by Kevin Lu that let people build apps across web, mobile, and Slack from natural language, and had pivoted into an agent platform doing browsing and integrations.TechCrunch reported the deal on July 7; both Bud and Orchids were set to shut down by July 18. The acquisition slots into Figma Make and the company’s Codex and Claude Code integrations, and the direction is unmistakable.Figma is buying its way from being a design tool to being a design-to-software platform, which is a polite way of saying it is trying to own the pipe that used to run through designers and engineers as separate humans.I want to be honest that watching this happen produces a strange feeling, because the same company that gave designers their most beloved tool is now systematically removing the reasons a company needs as many of them. That is not villainy, it is just where the value moved, and Figma would rather be the one moving it than the one it moves past.A company whose moat is “designers love us and know us” reads the Config announcements correctly as a warning that loving and knowing a tool is a depreciating asset, so it is racing to reattach itself to the part of the work that is still scarce, which is the production of working software from an intent.There is one detail in the Bud story that I think is the most important thing in either announcement, and it got buried. BBC reporting had flagged security vulnerabilities in apps built with Orchids. Sit with that. The pitch of vibe-coding into production is that anyone can now ship software, and the quiet asterisk is that a lot of the software being shipped is broken in ways the person who prompted it cannot see, because they never learned what a vulnerability looks like.The file became a runtime, and the runtime ships to real users, and nobody in the loop necessarily knows what they are doing. The moat did not disappear so much as relocate to a place almost nobody is standing yet, which is the judgment about whether the generated thing is actually correct and safe and good.What “knows Figma” was secretly a proxy forThe optimistic read, and I do hold it about half the time, is that “knows Figma” was never really the skill anyone wanted to hire. It was a proxy. When you hired for tool fluency, you were using it as a cheap, legible signal for something harder to measure, which was whether the person had judgment about interfaces.The tool test was a decent filter, because in the old world you could not produce a good interface without first mastering the tool, so tool mastery was correlated with the thing you actually cared about even though it was not the thing itself.The tools just severed that correlation. You can now produce a plausible interface without any tool mastery at all, which means tool mastery has stopped being evidence of judgment. And that is good news for the people who had the judgment and bad news for the people who had only the fluency, which turns out to be a lot of people, because the fluency was where the visible effort went and the judgment was the part you could coast on as long as the fluency was carrying you.This maps onto something Jack Cheng wrote in his essay for Every about what stays valuable when creative work gets easy: when technology changes what is difficult, what counts as skilled changes with it. Figma just changed what is difficult. The difficulty used to be making the thing in the tool. The difficulty now is knowing which thing is worth making and whether the generated version is any good, and those were always the harder skills, they were just hidden behind the tool difficulty for so long that people forgot they were separate.So the résumé line dies and something more demanding replaces it. Not “knows Figma,” but something closer to “has opinions about interfaces that are correct often enough to bet money on,” which does not fit as cleanly on a LinkedIn skills section and cannot be verified with a take-home test. The hiring problem gets harder in exactly the way the work got harder, because the easy proxy stopped working and now you have to evaluate the actual thing.We have watched this exact movie in other roomsThe reason I am fairly sure about the direction, even when the specifics are murky, is that the dissolution of a tool moat is not a new event, it just has a new subject. Every production craft that was ever hard has gone through this, and the sequence rarely varies. The skill of operating the hard tool is scarce, careers get built on the scarcity, then the tool gets easy and the scarcity evaporates and the value relocates to whatever the easy tool cannot supply.The darkroom technician who could coax a usable print out of a difficult negative was indispensable until the print became a slider. The typesetter who knew the physical craft of setting lead was a whole profession until desktop publishing put the capability on every desk. In each case the people who mistook their tool mastery for their value got blindsided, and the people who had actual judgment about the output did fine, because judgment was the part the easy tool never touched.What Config did is run that sequence for the design file at unusual speed. Normally a tool gets easy over years, and there is time for the incumbents to notice and adjust, to retrain, to move up into the judgment work before the floor gives way. Generative Plugins compressed the “tool gets easy” step into a single keynote, and the Bud acquisition compressed the “the platform absorbs the whole pipeline” step into the two weeks after it.The speed is the cruel part, because the people whose value was tool fluency do not get the gradual warning that earlier crafts got. They got a June keynote and a July acquisition, and the résumé line that meant something in the spring means noticeably less by late summer.There is a wrinkle specific to this round that the darkroom and the typesetter never had to deal with, and it is the security detail buried in the Bud story. When the print got easy or the type got easy, a bad print was just an ugly print, visible and self-correcting, the kind of failure the maker could see and fix. When the software gets easy, a bad build can be broken in ways completely invisible to the person who prompted it, which is exactly what the flagged vulnerabilities in Orchids-built apps were a preview of.So the relocation of value is not just toward “which output looks good,” it is toward “which output is actually correct and safe,” and that is a harder and less visible judgment than the aesthetic one. Almost nobody in the new easy-tool pipeline is positioned to make it, because the easy tool actively encourages people to skip the expertise that would let them.That is the piece I would put money on. The value did not evaporate, it moved to the judgment about correctness and quality that sits downstream of the now-free production, and that judgment is scarcer than the tool fluency ever was, because tool fluency at least had a tutorial and this does not. Which means the interesting hires, going forward, are not the people who can drive the tool fastest.They are the people who can look at what the tool produced and know, quickly and correctly, whether it is any good and whether it is safe to ship, and can explain the answer to a room that no longer contains anyone else who can tell.The uncomfortable part for people who are good at the toolI keep meeting designers who are deeply good at Figma, and I do not know how to feel about what is coming for them, so I am not going to pretend the arc bends cleanly upward for everyone. Some of these people are good at the tool and also good at the judgment, and they are going to be fine, because the tool getting easy just multiplies their judgment across more output.But some of them are good at the tool and average at the judgment, and they built a real, respectable, well-compensated career on being the fastest and most fluent hands in the room, and the room no longer pays a premium for fast, fluent hands. That is a loss for real people, and the industry’s habit of narrating every disruption as a net positive for everyone is its own small dishonesty.The advice that follows is not comfortable and I would rather it be true than comfortable. If your value is that you know the tool, you are holding a depreciating asset, and the depreciation just accelerated.The move is to spend the time the tool is about to give you back on the part it cannot do, which is developing an actual point of view about what good looks like and why, the kind you can defend when an agent hands you fifty plausible options and you have to know which one is right and be able to say why.Figma’s own tagline, “no tool should limit where an idea can go,” is accidentally the whole point. The tool stopped being the limit. Which means the limit is now you, specifically the quality of your ideas about what should exist, and that is a much scarier place for the limit to live than inside a piece of software you could always just study harder.The keynote was pitched as liberation and the acquisition was pitched as ambition, and both framings are true and both are also a redistribution. The value that used to sit in knowing the tool is moving to knowing what to do with it, and the people who spent their careers on the first thing are being quietly asked to have spent them on the second.A lot of them did. Some of them did not, and found out at Config that the thing they were best at is the thing the tool now does for free.This story is published on Generative AI. Connect with us on LinkedIn and follow Zeniteq to stay in the loop with the latest AI stories.Subscribe to our newsletter and YouTube channel to stay updated with the latest news and updates on generative AI. Let’s shape the future of AI together!When the Plugin Is a Prompt, Knowing Figma Is No Longer Enough was originally published in Generative AI on Medium, where people are continuing the conversation by highlighting and responding to this story.Source: Generative AI Pub — Published — Category: Image AI