When a developer at Jane Street wrote that he now "designs more with Claude than with Figma," the post racked up 225 upvotes on Hacker News within 48 hours, along with 209 comments. Jane Street is one of the largest quantitative trading firms in the world. So when one of its engineers describes building screens inside an AI chat window instead of Figma, the story reads as more than a convenience anecdote — it signals that a shift in working tools is already underway. Tellingly, many of the people who responded were in the same position: stuck handling design alone, with no designer to lean on.
Building Screens with Prompts
His workflow goes like this. He describes the screen he wants in plain language — layout, colors, interaction patterns, even component structure — and Claude generates the HTML, CSS, and React code. He checks the result instantly in a browser, and if something is off, he asks for changes. The Figma routine of setting up frames, arranging components, and preparing developer handoff files gets replaced by a loop of typing text and reviewing code.
The biggest difference he noticed was the switching cost. Working in Figma, he constantly toggled between design mode and development mode, and translating a drawn UI into code introduced unintended gaps. With Claude, that gap never exists in the first place: describing the screen and generating the code happen in the same act. He summed up the experience as "the feedback loop changed."
In the Hacker News comments, people in similar situations kept chiming in. Backend and full-stack developers who "have no designer but still have to ship UI" voiced their agreement. A recurring opinion was that making specific requests to an AI beats spending time learning Figma. Many noted the approach works especially well for internal tools and admin dashboards — places where functional completeness matters more than visual polish.
A Good Time to Ask Whether Figma Is the Right Tool
There's a story that keeps surfacing in Korean solo-founder communities: people pay for a Figma subscription but only ever use it for wireframes and rough mockups. Figma Professional runs $15 a month. Paying that rate while barely touching the collaboration or prototyping features is a pattern that repeats across many productivity tools.
One of the main reasons solo founders and solo PMs use Figma is "to draft something before handing it to a designer." But if there's no outsourced designer or in-house design team, that Figma mockup isn't an intermediate file passed to a developer — it's a final deliverable you have to implement yourself. In that situation, it's worth asking whether Figma still earns its place.
There's a recurring pattern in how technical tools become accessible: the tool itself doesn't get simpler — the interface to it changes. Aerial photography, data visualization, and 3D printing all followed this arc. Drone control software was once the domain of professional pilots, until a new interface put the same equipment in the hands of videographers and farmers. AI coding assistants are headed the same direction. That doesn't mean they will fully displace the incumbent tools, though.
Why This Approach Doesn't Work for Everyone
There is clear skepticism about this trend, too. A good share of the Hacker News comments offered concrete counterexamples. The most frequent complaint: "UI that Claude builds is functional but inconsistent." Commenters described button spacing that drifted subtly between screens, color tones that didn't match across pages, and responsive layouts that broke on specific views. The pattern they reported: fast at producing a single screen, but labor-intensive when stitching many screens into one coherent system.
UX practitioners pointed to a more structural problem. Conversing with Claude works well for revising the screen in front of you, but it's poorly suited to designing flows with the entire user journey in view. It responds quickly to "change this button color," but hits its limits with "structurally fix why users drop out of this flow." Screen-level edits and experience design are work at different altitudes.
It also matters that the approach worked for the Jane Street developer because he could write UI code himself. You need the ability to review generated code, spot problems, and steer the fixes before Claude's output becomes genuinely usable. A product manager or marketer with no development experience working the same way is unlikely to get the same results. Writing good prompts and judging the quality of the output are different skills.
A Way to Test This Right Now
Even so, there's a practical takeaway from this experiment: instead of framing Figma and Claude as rivals, reallocate which tool you use at which stage.
Early in a project, before the screen structure has settled, Claude can be faster. Describe what you want in specifics — "a pre-payment order confirmation screen, mobile-first, with product name, quantity, total, and a CTA button; white background, indigo accent" — and you'll get a draft faster than ten minutes of staring at a blank Figma canvas. You can collect team feedback on that draft, or move into Figma from that point on.
On the other hand, if you already have a design system, or your product depends on consistency across many screens, you're better off keeping Figma's defined style guide than using Claude's components as-is. Claude makes fresh styling judgments with every request, and the longer a project runs, the more those differences compound.
The subscription math is worth running, too. Figma Professional is $15 a month; Claude Pro is $20. If you're at an early stage where paying for both makes no sense, one option is to pause the Figma subscription, rough out your initial screens with Claude, and switch over once the product takes concrete shape.
Count how many of the Figma files you created in the past month actually made it into development — and whether the time spent on them went toward sharpening the idea or operating the tool. I'd argue that's the audit to run before choosing any tool.
