What Wektor actually does with your data

What stays on your machine, always

• Your uploaded CSV/XLSX files, and the full contents of any database query you run, are processed entirely locally by the Python engine on your computer. They are never uploaded anywhere by Wektor.
• All chart data, computed metrics, and generated reports/PDFs/slides are produced locally and stay on your machine unless you choose to share them yourself.
• Your AI provider API key is stored using your OS's native credential store (Windows Credential Manager / macOS Keychain / Linux Secret Service) via the keyring library, or, if none is available, in a locally Fernet-encrypted file. It is never logged, and the UI only ever shows a masked version.

What is sent to your chosen AI provider

Wektor is Bring-Your-Own-Key: you provide your own API key, and requests go directly from your machine to that provider using your key — Wektor's own servers are never in this path (Wektor doesn't operate a server at all). For each analysis or chat request, the following is sent to the model you've configured:

• Column names, coarse data types, and aggregate statistics — e.g. a numeric column's min/max/mean, a categorical column's number of unique values, a date column's date range, and the % of missing values. These are derived from your data but are not raw records.
• Your natural-language question, when you use the chat feature.
• The row count of your dataset (a plain integer).

Actual row-level data is never sent to the AI model. The model only ever writes Python analysis code based on the schema/stats above; that code is then executed locally, against your real DataFrame, inside a sandboxed process — so results are computed from your real data without your real data ever leaving your machine.

What Wektor does not control: your AI provider's own data retention and model-training policies for API traffic. Check your provider's terms directly if that matters to you.

Database Connector & Daily Watch

• Only a single read-only SELECT (or WITH ... SELECT) query is ever executed — INSERT/UPDATE/DELETE/DROP/etc. are blocked before the query reaches your database.
• Queries are subject to a timeout and a result-size cap, enforced locally.
• Query results are exported to a temporary local CSV so they can flow through the same analysis pipeline as an uploaded file. These temp files are automatically cleaned up (anything older than 24 hours is deleted on the next app startup) and are never sent anywhere.
• Your database password is stored the same way as your AI provider API key — never in plain text.

Local storage summary

Telemetry

Wektor does not currently collect or transmit any usage analytics, crash reports, or telemetry of any kind. There is nothing to opt in or out of because nothing is being sent.