Add a web page with a crawl

A web crawl reads a public web page and brings its text into Chat, so an answer can be grounded in it and cite it the same way it cites a document you uploaded. It’s the quickest way to add a supplier’s spec page, a standard, or a competitor’s pricing page to your material without downloading and re-uploading anything.

There are three ways to start one.

Add a page to a collection

Open a collection, choose Add Source, and paste the URL of the page you want. Chat fetches the page, extracts its text, and adds it to the collection as a document — from then on it behaves like any other document in that collection, and counts toward the collection’s document limit.

This is the option to use when the page is reference material you’ll come back to. Everyone who can see the collection can use it.

Paste a URL into a message

Include a URL in a message and Chat will offer to read the page as part of answering you. Nothing is added to a collection — it’s used for that conversation only.

Ask for it in plain language

You can simply ask: “Crawl https://example.com/pricing and summarize their plans.” The assistant fetches the page and works from what it finds.

Public pages only

A crawl reaches pages the way a browser would, without signing in. Anything behind a login, a paywall, or your own network won’t be reachable — upload the document instead, or connect SharePoint.

Crawled pages don’t refresh themselves

A crawl captures a page as it was at the moment you ran it. If the page changes later, the copy in your collection doesn’t change with it — run the crawl again to pick up the new version.

Related

• Create a collection and upload documents — limits and supported file types.
• Connect SharePoint — bring in documents from your SharePoint site.
• Attachments and web search — searching the web generally, rather than reading one specific page.