Privacy

Last updated 1 November 2026. Belgium Data Desk is operated by Vecture (vecture.be).

The website

Belgium Data Desk has no accounts, no advertising and no analytics or tracking scripts. Dataset profiling and SQL analysis run inside your browser; files you choose or drop stay on your device. Your pins and preferences are kept in your browser’s local storage and never sent to us.

The AI-assistant connector (MCP)

When an assistant calls the connector, it sends us the words of the search or the dataset id it asks about. We use them only to produce the answer and do not store them. The assistant decides what to send; the connector never reads your conversation, your files or the assistant’s memory. The connector does not require sign-in or set cookies. We keep aggregate service-usage counters such as which MCP tool was called, whether it succeeded, result count, response time and a coarse topic category (for example energy or mobility); these counters do not contain search text, dataset ids, IP addresses, identities or conversation content. We do not build user profiles. The assistant you use (for example ChatGPT or Claude) has its own privacy terms for your conversation.

When an assistant asks for a file preview (the preview_dataset tool), the connector reads the first 64 KiB of that dataset’s public CSV or TSV file from the publisher’s own server and returns the first rows. Only files already listed in the public catalogue can be read; the request to the publisher carries no information about you, and the file is not stored.

Retention

Search words and dataset ids are not kept after the answer is sent. The aggregate usage counters described above are kept for up to three months in Cloudflare’s analytics storage and are then deleted automatically.

Sharing

We do not sell, rent or share data with anyone, and there is no advertising. The only third party involved is our hosting provider, below.

Our hosting provider

The service runs on Cloudflare Workers. Like any web host, Cloudflare processes standard request information (such as IP address and request time) to deliver the service and keep it secure, under its own privacy terms. We do not receive or keep this information as a log of individual users.

Fetching public files

For some publisher files the site offers a bounded relay so your browser can read them. The relay only fetches files that are already listed in the public catalogue, forwards no cookies or personal headers to the publisher, and does not store the files.

Personal data in the catalogue

The catalogue contains publisher contact points (for example a department’s email address) exactly as the publishers publish them in open-data catalogues. If you believe a contact detail should not be shown, tell the publisher or contact us and we will remove it from our copy.

Contact

info@vecture.be