Turn any website into a clean list

This is the one you'll use most. Most of the time you don't need anything clever. You grab a web page, you hand it to Claude, and you ask for the parts you want in plain words. It reads the page and gives you back a tidy list. That's the whole thing. This is AI web scraping at its most basic, and it covers more of the web than you'd think.

The one wall everyone hits first: Claude can't open websites on its own. So you bring the website to it. Fetch the page, paste it in, and ask.

The shape, in three steps

1. Grab the page

This is a single, simple request that downloads the page's contents, the same thing your browser does when you visit. It handles the large majority of the web. Save a copy of what you grabbed, always, because the page that breaks your work two months from now is the one you forgot to keep.

2. Trim the junk

A web page is mostly clutter you don't care about (menus, ads, styling code). You strip that out so what's left is just the meat. This makes the next step cheaper and sharper. You don't have to be precise, just toss the obvious noise.

3. Ask Claude for what you want

Here's the part that changed everything. Instead of writing fragile instructions like "grab the third box from the top" and praying the next site is laid out the same way, you just describe what you want and let Claude find it.

Why this beats the old way: every website is built differently. A rigid scraper works on the first dozen sites and breaks on the next dozen, because someone used a different template. Claude reads the page the way a person does and pulls the right thing by what it means, not where it sits. One instruction, a thousand layouts.

When you grab a page and it comes back nearly empty, that's your cue the page needs a real browser to load first. When even that gets turned away, you go through the back door. But you reach for those last, because most of the web answers a simple request just fine.