Importing content from an existing website

If you already have a website, you don't have to retype its content into a document. On the same step where you'd upload a file, you can paste the address of a site you own and let A3 iFrame read it for you. It collects the text and images it finds and hands them to the AI as the source for your new site.

This sits alongside the other two ways to start, described in building your first website, step by step: uploading a document, or describing your site with AI.

You have to own the site you import

Importing is only available for domains you can prove you control. This protects other people's content, and it isn't optional — nothing is fetched from a host until a proof is in place. You have two ways to prove it:

• A DNS record — add the TXT record we give you to your domain's DNS. This covers the name you prove and everything beneath it, so proving example.com also covers blog.example.com.
• A file on your site — put a small file at the exact path we give you. This covers only the exact host you proved, and nothing else.

The screen shows you the precise record or file to create, with copy buttons, and re-checks when you ask it to. DNS changes take a little while to spread, so if the first check doesn't pass, wait a few minutes and check again. If DNS is unfamiliar, DNS records explained covers the basics.

One page, or a crawl

Once your domain is verified you choose how much to read:

• Single page — reads just the address you gave. Good when one page, such as an About or a home page, already says everything.
• Crawl — follows links within the same site and reads up to 20 pages. Good for a multi-section business site.

A crawl runs in the background and shows its progress as it goes, including the page it's currently reading. You can stop it at any time, and stopping is not a failure: whatever was collected before you stopped is kept and imported.

What gets collected

The importer takes the readable text of each page, plus up to 4 images per page in PNG, JPEG, WebP, or GIF format. Very small images are skipped, because they're usually icons and spacers rather than content. Everything else — scripts, styling, layout — is deliberately ignored, because the AI designs your new site from scratch rather than copying your old one.

The limits, so nothing hangs

So that an import always finishes in a predictable time, a few bounds apply:

• Up to 20 pages in a crawl, and around 2 MB of content per page.
• Roughly 10 seconds to fetch any single page, and about 2 minutes for a whole import.
• Up to 20 imports an hour for your account.

We also respect your site's robots.txt. Our importer identifies itself as A3iFrameImportBot, so if your own crawl rules block it, allow that name and try again.

The four outcomes

An import ends in one of four ways, and the screen tells you which:

• Succeeded — content was collected and is ready to generate from.
• No change — the fetch worked, but there wasn't enough usable text to build from. Try a content-rich page, or switch to a crawl.
• Refused — the address wasn't allowed, usually because the domain isn't verified or crawl rules block us.
• Failed — the site couldn't be reached, or it timed out.

Which plans include it

Importing from a URL is available on the paid plans. On the Free plan you can still build a complete site by uploading a document or describing it with AI. See plans and what's included.

When the import finishes, carry on exactly as you would with an uploaded document: set your design preferences and generate. If an import won't verify or keeps failing, open a ticket from support with the address you're trying and we'll take a look.