Claude has eyes and almost nobody uses them. The second you start sending it screenshots, it changes what you can build. Here are the two ways I use it every week.
Most people miss this. Claude can see images. Drop a screenshot, a photo or a design into the chat and it reads it like you would, from the layout and colors to the spacing and the vibe. If you only ever type words at it, you’re using half the tool.
Once you start feeding it pictures, two superpowers open up, especially if you build anything visual like a website, an app or social posts.
Where This Works
In the Claude app, drag an image into the message box or paste it (Cmd/Ctrl + V). On your phone, attach it like a photo. In Claude Code in the terminal, drag the image in, paste it with Ctrl + V (yes, even on a Mac; it’s Alt + V on Windows), or give it the file path. Then talk about what you sent.
When you’re building something visual, stop trying to describe the look in words. "Make it modern, clean and premium" means nothing, and every AI builds a different bland version of it. Show it instead.
01
Make it look like this
Find a website, app or post whose design you love. Screenshot it, drop it into Claude, and say "make it look like this." It pulls the actual design patterns from the image, so what you get back feels custom instead of like every other AI-generated page on the internet.
02
Borrow from another industry
Pull inspiration from a totally different industry, not a direct competitor. If you sell skincare, screenshot a beautiful architecture studio or a coffee brand. That’s how you end up looking like nobody else instead of a copy of the brand next door.
03
Treat it as a starting template
The screenshot gets you most of the way past the blank page. Then put your own spin on it so it’s yours.
Pro Move
Drop in two or three screenshots at once and say "take the layout from the first, the color feeling from the second, and the type from the third." Mixing references is how you get a look that feels original instead of borrowed.
This is how I build now. When Claude makes something visual, don’t take the first version and hope it looks good. Make it look at its own work the way a person would, and fix what’s off before it ever reaches you.
01
Build, screenshot, critique, fix, repeat
I tell Claude to build the thing, then screenshot the actual result, then judge that screenshot. Is anything overlapping, off-center or too cramped? It spots the problems, fixes them, screenshots again, and repeats until it’s genuinely good.
02
Get a better first draft
By the time it hands me a "first" version, it has already done three or four rounds of building and checking itself. The obvious stuff gets caught before it lands on my screen, so I stop being the one who finds every little visual bug.
The Honest Setup Note
For Claude to screenshot a page on its own, it needs a way to open it. The easiest is the Code tab in the Claude desktop app, which can run your app in a built-in browser pane and take screenshots to check its own changes (how preview works). In the terminal, connect Claude Code to Chrome with the Claude in Chrome extension and start it with claude --chrome (Chrome setup). No setup yet? Run the loop by hand. You take the screenshot, paste it back, and say "here’s how it actually looks, what’s off? Fix it." Same result, you’re just the camera.
The Real Win
Once Claude has eyes, it stops only following instructions and starts judging its own output. Show it what good looks like, and make it check itself against that. That one shift is the difference between "AI slop" and something you’re proud to put your name on.