What it fetches

The home page and at most three more pages, usually contact and about. It looks for what you published in order to be contacted: a company email address and the organisation number. A personal identity number is never stored. Sometimes it visits the home page once more to see how the site is built: which platform it runs on, and whether it has analytics, a contact form and a chat. The home-page text is also read by an AI model that writes a short summary of what you offer and when you are open; the summary is stored, not the text. It also checks whether the site has an llms.txt, the file AI search reads; only whether the file exists is stored, not its content.

How it behaves

It reads your robots.txt first and follows it. It fetches one page at a time, with at least one second between requests to the same site, and only HTML: at most 1 MB and 10 seconds per page. It does not fill in forms, log in or try to get past blocks. If the server says no, it gives up. Email addresses hidden by Cloudflare's email protection are not decoded.

Two more visits you may see in your logs

We measure how fast the home page loads with Google PageSpeed, which loads the page once from Google's servers. Companies we are considering contacting may also get a review of up to ten pages, made with the tool OpenSEO. Its crawler follows robots.txt. If your robots.txt says no to KaftsmallBot, we make neither measurement, and before the review starts we read your robots.txt for KaftsmallBot once more.

How to stop it

Put this in your robots.txt and it will not come back:

User-agent: KaftsmallBot
Disallow: /

Or email sebastian@kaftsmallonline.se and we will block you in our register by hand. Neither the robot nor the review will then visit your site again, and you get no email from us.

Why it exists

We build websites, run ads and set up AI chats for small businesses. The robot helps us find companies where we think we can be useful, so we can get in touch with something specific instead of a mass mailing.