Stagewise Wants You to Point at a Button and Say 'Make It Green'

The Macro: The Gap Between What You See and What You Code

Frontend development has a fundamental workflow problem that nobody talks about because developers have just accepted it. You look at your app in the browser. You see something wrong. A button is in the wrong place, a color is off, the spacing looks weird. Then you leave the browser, open your editor, find the right file among hundreds, locate the right component among dozens, change a value, save, switch back to the browser, wait for the hot reload, and check if it worked. Repeat forty times a day.

That loop is absurd when you say it out loud. You are looking directly at the thing you want to change, but you cannot touch it. You have to go somewhere else, translate the visual problem into a code problem, solve it there, and come back. It is like editing a photo by writing PostScript by hand.

The frontend development tools market is massive and competitive. Cursor, Windsurf, and Claude Code are all making AI-assisted coding faster inside the editor. Vercel’s v0 generates UI components from prompts. Lovable and Bolt build entire applications from descriptions. But all of these tools operate at the code level, not the visual level. You still describe what you want in words or code. You do not point at it.

This is not a new observation. Browser devtools have let you inspect and tweak CSS for years. But those changes are ephemeral. They vanish on reload. And they only handle styles, not component logic or layout restructuring. The dream of “click on something in your app and change it for real” has been floating around frontend development for a decade without anyone shipping a production-quality version of it.

The open-source AI tooling space is also crowded right now. Over 4,800 GitHub stars means something in terms of developer interest, but developer interest and daily active usage are different metrics. The question for any open-source dev tool is whether it becomes part of the workflow or gets starred and forgotten.

The Micro: A Browser Toolbar That Rewrites Your Codebase

Glenn Tows and Julian Goetze built stagewise out of Germany as part of Y Combinator’s Summer 2025 batch. The team is two people. The product has accumulated over 4,800 GitHub stars and 130,000 downloads, which is significant traction for a developer tool at this stage.

The core product is an open-source browser toolbar that overlays your locally running application. You click on any element in your app, type a natural language instruction like “make this green and move it to the top right,” and the agent modifies the actual source code in your codebase. Not a temporary CSS override. A real code change.

That distinction matters enormously. Every frontend developer has used browser devtools to prototype a visual change, liked the result, and then spent fifteen minutes trying to figure out which CSS file to edit to make it permanent. Stagewise collapses that entire loop into one step.

The toolbar is framework-agnostic, which means it works with React, Vue, Svelte, or whatever your production stack uses. It automatically collects browser context and application metadata so the AI agent understands not just the DOM element you clicked but the component hierarchy, the styling system, and the surrounding code. That context is what separates a useful agent from one that makes random edits and breaks your layout.

The newer version of the product goes further. They have built what they call a purpose-built browser for developers with a coding agent baked in. It gives the agent access to the console and debugger across all browser tabs. You can make temporary changes to test an idea or permanent edits that go directly into your codebase. There is a reverse engineering mode that lets you extract components, styling systems, and color palettes from any website. IDE integration is optional, showing relevant files and modifications in your preferred editor.

The reverse engineering angle is clever and I have not seen it in competing products. Being able to visit a website you admire, extract its design system, and apply those patterns to your own project is the kind of workflow that designers and product managers would actually use. It lowers the barrier between “I want my app to look like that” and actually making it happen.

The product is currently in waitlist mode for the browser version, with the open-source toolbar available now. Sales inquiries go to the team directly.

The Verdict

Stagewise is solving a real friction point in frontend development that I have personally experienced hundreds of times. The visual-to-code loop is genuinely broken and this is the most direct fix I have seen for it.

What I want to know at 30 days is how well the agent handles complex component structures. Clicking a button and changing its color is the demo. Clicking a card component and saying “restructure this to show the price more prominently” is the real test. The harder the structural change, the more context the agent needs, and context windows have limits.

At 60 days, the question is adoption depth. Are developers using this daily or is it a novelty they reach for occasionally? The 130,000 downloads suggest strong initial interest. Retention is the real metric. At 90 days, monetization. Open-source dev tools with strong traction have a clear path to commercial products, and the browser-based version behind the waitlist is presumably where the revenue model lives.

The two-person team from Germany building an open-source tool with nearly 5K stars and 130K downloads is doing something right. I think stagewise has a legitimate shot at becoming a standard part of the frontend development toolkit, especially for teams where designers and product managers want to make UI changes without filing tickets. That alone is a large enough use case to build a company around.