KairossBot: the Kaiross page reader
If you find the user agent below in your site logs, a person using Kaiross has asked an agent to read one of your public pages. Here you find how it behaves, how to block it and whom to write to.
Last updated:
KairossBot/1.0 (+https://kaiross.ch/bot) What it is
Kaiross is an AI agent platform for companies. With the Cowork feature a user gives a task to an agent, for example “prepare a list of companies in a sector with their public company contacts”. To do this the agent searches the web and reads some public pages. KairossBot is the name it uses when it reads a page. It is not a crawler that indexes the web: it reads a few pages for a specific task and does not build a general archive of websites.
Who starts it
Every read comes from a task started by a Kaiross customer, who decides its purpose. Kaiross provides the platform and sets the technical rules described on this page. The results (for example an Excel file with a list of companies) belong to the customer who started the task, who is responsible for them as data controller when they contain personal data.
How it behaves
The rules are fixed and apply to all customers:
- It respects the robots.txt file of every site. If robots.txt cannot be reached, the page is not read.
- It reads only public pages with GET requests over http or https. It does not fill in forms, does not log in, does not enter restricted areas and does not send data to your site.
- It has limits per task: a maximum number of pages and megabytes, at most 15 pages per site and 5 MB per page.
- It reads text, HTML, PDF and CSV. It opens only addresses written in the task, found in search results or present as links on pages already read.
- Instructions hidden in pages are never executed: the content of a page is treated as data, not as an order.
- Every page read or blocked is recorded in the task log, visible to the customer.
How to block it
Add these two lines to your site’s robots.txt file. You can also disallow only some folders, with the usual robots.txt rules. If the file has no group for KairossBot, the rules of User-agent: * apply.
User-agent: KairossBot
Disallow: / Which data it reads
The visible text of the public pages the task needs: usually the company name, address, website, and the company email and phone published by the company itself. Tasks, log and files are stored in Zurich in the customer’s space and are deleted according to the retention the customer sets. The texts go through the AI provider chosen by the customer.
Removal and objection
If you do not want your data to appear in the results of a Kaiross customer, write to info@kaiross.ch stating the site or page and the data concerned. We forward the request to the customer who started the task, who is the data controller, as long as the task log is kept. If a customer has contacted you, you can also contact them directly: they must tell you where they got your data and respect your objection. You can also lodge a complaint with the FDPIC or, in the EU, with the supervisory authority of your country.
Contact
To report unusual behaviour by KairossBot on your site, write to info@kaiross.ch, if possible with date, time and page address. Platform operator: Neuvest Mancuso, Via Carlo Maderno 19, 6900 Lugano, Switzerland.