Privacy statement
Last updated on October 13, 2026
This is a translation. The Dutch version of this page is binding; if the two differ, the Dutch text prevails.
LexAgent gives you and your AI assistant access to the official Dutch and European legal sources. We process as little personal data as possible: your documents stay in your own browser, and of what you look up we only keep that there was a request, not what you searched for. Below is exactly what we do process.
Who we are
LexAgent is a product of SWIS B.V. in Leiden, the maker of Vragen.ai. SWIS is the controller for the processing of the data in this statement. If you use LexAgent through your organization, we set out the arrangements for that in the agreement with that organization.
Your documents stay in your browser
Dossiers, documents, findings and drafts in the Workspace are kept in your own browser's storage. They are not uploaded to our servers and we cannot access them. If you clear your browser data, they are gone, so make an export if you want to keep them.
Before the assistant may read a document, you review the personal data in it. After that, the assistant only gets a version in which names, addresses and other details have been replaced by labels such as [Persoon 1].
What data we process
| Data | Purpose | Legal basis | Retention period |
|---|---|---|---|
| Your account: name, email address, password (stored encrypted), your organization and which sources you may use. For the free version, also the end date and your credits. | Letting you sign in and giving you the right access. | Performance of the agreement | As long as your account exists. We delete a free trial account 30 days after the end of the trial period. |
| Requests to the sources: time, which tool (for example "search legislation"), the name of your app (such as Claude or ChatGPT), duration, result and size. Not your search terms and not the texts. | Keeping the service running, detecting outages and misuse, and being able to show usage per organization. | Legitimate interest | 30 days |
| Counts in the free version and with managed models: how many questions you asked, how many claims you had checked and supported with sources. No content. | Keeping track of credits and billing usage to your organization. | Performance of the agreement | As long as needed for our records. After your account is deleted, no longer linked to you. |
| Contact requests from the Workspace: name, organization, email address, the reason, and counts of what you did in the Workspace. No content. | Getting in touch with you. | Consent | 6 months |
| Technical data: IP address in server logs, and error reports without the content of your requests. | Security and fixing errors. | Legitimate interest | Briefly, until the error is fixed or the logs have been rotated. |
The assistant in the Workspace
The Workspace works with a language model. The route your question takes depends on how you set up the model:
- With your own key (Anthropic or OpenAI), the request goes directly from your browser to that provider. We do not see it. That processing is governed by the provider's terms.
- In the free version and with a model through LexAgent, the request goes through our server to Azure OpenAI in the European Union. It only passes through our server in memory; we do not store the content and ask the provider not to store it either. For each question we record the usage, linked to a number of your account, not to your name or email address.
LexAgent in your own AI assistant
If you connect LexAgent to Claude, ChatGPT or another assistant, you sign in with your LexAgent account and give that app access. Your conversation with the assistant is covered by that provider's privacy statement. We only receive the search and read requests the assistant makes to LexAgent. We use them to answer; of those requests we only keep the data listed in the table above. We store retrieved public sources temporarily, not linked to who asked for them, so that the next question is answered faster. You can revoke the connection in your assistant at any time.
Who else processes the data
- TrueFullstaq (Amsterdam): hosting of LexAgent. The data stays in the EU.
- Microsoft Azure OpenAI (EU): the language model in the free version and with a model through LexAgent.
- Sentry (EU): error reports, without the content of your requests.
We have data processing agreements with these parties. We do not sell data and do not use it for advertising.
Cookies
LexAgent only uses cookies that are needed for it to work: to keep you signed in (and, if you choose so, to keep you signed in longer) and to secure forms. Your choice of a light or dark theme is stored in your browser. We do not use analytics or advertising cookies. The question box on the home page is a Vragen.ai widget; the Vragen.ai privacy statement applies to it.
Security
We protect data with appropriate technical and organizational measures. SWIS is certified to ISO 27001 and NEN 7510. Connections are encrypted, and only employees who need it for their work can access the data.
Your rights
You can access your data, have it corrected or deleted, get a copy, object to a processing activity or withdraw your consent. Send your request to info@swis.nl; we respond within four weeks. The data protection officer (functionaris gegevensbescherming) of SWIS is Björn Brala (bjorn@swis.nl). If you disagree with how we handle your data, you can file a complaint with the Dutch Data Protection Authority (Autoriteit Persoonsgegevens).
Changes
If anything changes in how we process data, we update this statement. If it is a significant change, we will let you know.