Your client wants the headline bigger, the CTA in their brand blue, and the subhead rewritten. Three changes, three prompts, three waits, three re-checks. As of September 3, 2026, that's a toolbar.
What is Visual Edits?
Visual Edits is a click-and-change editing layer in Bolt.new, an AI app builder from StackBlitz. You click an element in your running app, change it with a toolbar, stack up as many changes as you want, and send them to the model in one batch. As of September 3, 2026, it's available to every Bolt.new user on both free and paid plans.
How does Visual Edits work?
Visual Edits lives on the Select button you already use, now marked new. Click it and a short walkthrough runs the first time. Dismiss it and it stays dismissed in that browser.
Selecting works the way it always has. What changed is what happens after you select. Every element gets a toolbar. Elements with text in them get a bigger one: font size by the pixel or from a size picker, styling, lists, links, and editing the copy in place with the keyboard shortcuts you already know.
That toolbar's color picker reads the palette off your rendered page, so the colors it offers are the colors your app is already using.
Selected elements now carry plain-language names, not div and h1.
Why do the changes batch?
Each change you approve stacks into a list. Keep going. Change your mind and adjust an element again, and it updates that change rather than stacking a second one. Discard one and the preview snaps back.
When you like what you see, save. The whole stack goes to the model at once.
Batching is what makes this quicker than prompting change by change. One instruction set, one round trip, and the model gets exact values instead of a description it has to interpret.
Worth knowing before you click save: that batch goes to your currently selected model, so it spends tokens the way a prompt does. Less than describing each change in chat, but not free.
What Visual Edits doesn't do
This release covers text and styling. You can't drag elements around or rearrange a layout, and applying one change across many elements at once isn't something to count on yet. The prompt box is still there for everything the toolbar doesn't reach. The scope is narrow by design.
It pairs well with decks built in Bolt Slides, where most of what you're adjusting is text.
Where this pays off
The revision cycle is where agency margin goes to die. The client says smaller, you prompt, you wait, you check, they say actually bigger. Visual Edits turns that into a live session: make the changes while they watch, stack them, save once.
Only 4% of marketers use AI to write entire pieces on its own, and the norm is human-directed, AI-assisted work (HubSpot AI Trends for Marketers 2025). You still decide what on-brand means. Visual Edits stops you from describing it in a sentence.
