Subprocessors
A list of everyone we pass customer data to when operating Kóta. We announce a change to office administrators by e-mail at least 30 days in advance; the data processing agreement describes how to object to it.
This is a translation. If it differs from the Czech version, the Czech version prevails.
1. Our own infrastructure
Files, extracted text, the vector index, conversations and the account database are stored on a server in the Czech Republic that we operate ourselves. We have no hosting provider today. If we move the service to a cloud, we will add the provider to this list with advance notice, and the data will stay in the European Union.
Text recognition from scans (OCR) runs on this server and the data goes nowhere in the process.
2. Sub-processors of content
| Who | What for | Which data | Where | Safeguards |
|---|---|---|---|---|
| OpenRouter, Inc. (USA) | routing requests to language models: answers to questions, data extraction, summaries, document drafts | the text of the question and conversation, the excerpts and pages of documents needed for the answer | USA | data processing agreement with standard contractual clauses; under its terms it does not store inputs and outputs unless the customer expressly asks for it (we do not), and does not use them for training |
| The model host selected through OpenRouter, currently Microsoft Azure for the default model | running the language model | the same | depends on the endpoint; neither the EU nor the USA can be guaranteed | on every request we require an endpoint with zero data retention and no training; no other endpoint serves the request |
| Google Cloud EMEA Limited (Ireland), Gemini API | computing vector representations of text for search | excerpts of document text and the text of questions | Google infrastructure in the EU and outside it, including the USA | Google’s data processing addendum; a paid service, data is not used to improve products; Google LLC states that it adheres to the EU–US Data Privacy Framework |
| Scaleway SAS (France) | sending the service’s e-mails: invitations, address confirmation, password reset, limit warnings | e-mail address, name, office name, message text | EU | data processing agreement, processing in the EU |
On Google: under its policies the paid Gemini API may retain inputs for 55 days solely to detect abuse. We therefore do not promise zero retention there.
3. Processors for our own purposes
| Who | What for | Which data | Where | Safeguards |
|---|---|---|---|---|
| PostHog Inc. (USA), PostHog Cloud EU | measuring the use of the service, according to your choice on the Cookies and storage page | pages viewed and clicks without text, device and browser type, IP address; with consent a recording of your path through the app with text and images hidden | EU (Frankfurt) | data processing agreement with standard contractual clauses |
4. Independent recipients
These services are not our processors. Data goes to them because you ask for the function yourself, and they process it under their own rules.
| Who | When | Which data |
|---|---|---|
| Google (sign-in with a Google account) | if you sign in with a Google account | Google passes us your name, e-mail and account identifier |
| Mapy.com (map tiles and address lookup) | when you view a project’s map | the project’s address and your device’s IP address, straight from the browser |
| Czech Office for Surveying, Mapping and Cadastre | when you look up a plot | the cadastral area and the plot number |
5. Providers you connect yourself
If you enter your own key for an artificial intelligence provider in the settings or allow outside assistants (such as Claude or ChatGPT), we pass data to them on your instruction and under your contract with them. They are not our subcontractors.
Send questions about the list to [email protected].
Effective from October 1, 2026