---
title: "AI agent · Karel Navrátil"
description: "How the AI agent on Karel Navrátil’s website works: the answers are generated by artificial intelligence, are not legal advice and may be inaccurate."
url: https://navratilkarel.com/en/ai-agent
language: en
alternate: https://navratilkarel.com/ai-agent
---

# How my AI agent works

This is an English translation of the [Czech original](https://navratilkarel.com/ai-agent). Czech law applies. If the two versions differ, the Czech version prevails.

Questions on the website are answered by an AI agent, not by me. Its answers are not legal advice and may be inaccurate. I will always confirm what applies myself, by e-mail.

## 1. Who you are talking to

In the field “Ask my AI agent about this site” under the frequently asked questions and in the chat in the Ask window, you are talking to an artificial intelligence system, not to me. The answers are generated automatically by the large language model Claude by Anthropic. When it is not available, a model that Cloudflare runs for me (Workers AI) answers. I do not read the conversations and I do not reply in the chat.

The agent answers only from the texts of this website: the frequently asked questions, articles, talks and the description of my services. It can recommend what to read or put you in touch with me.

## 2. It is not legal advice

The agent’s answers are not a legal service or advice on your matter. A conversation with the agent does not create an attorney-client relationship or a contract for the provision of legal services. To advise you, I need to know your situation: write to me at [advokat@navratilkarel.com](mailto:advokat@navratilkarel.com).

## 3. The agent can be wrong

Artificial intelligence can write information that is inaccurate, incomplete or untrue, even when it sounds certain. So do not rely on the agent’s answer, and check anything that matters to you with me by e-mail. Only what I write to you myself applies.

## 4. What happens to your question

Please do not enter personal or confidential information into the agent. The text of your question (in the chat, also at most seven previous messages of the same conversation) is sent through Cloudflare to Anthropic so that the model can answer. I do not store it on the server and I do not link it to you. Anthropic deletes it within 30 days at the latest and does not use it to train its models. The chat conversation stays only in your browser’s storage until you close the tab.

I am the controller of the data; Anthropic and Cloudflare are my processors. You will find the legal basis, the transfer to the USA and your rights in the [privacy policy](https://navratilkarel.com/en/privacy-policy) (sections 2, 3, 6 and 8). Last updated: 12 October 2026.

---

Mgr. Karel Navrátil, attorney-at-law (Czech Bar Association No. 21034, company ID 19839448), Korunní 810/104, 101 00 Prague 10 · advokat@navratilkarel.com  
For AI agents: https://navratilkarel.com/en/agent.md  
The texts are general information, not legal advice on a specific matter.
