Station app privacy
Station runs on hardware you control. The short version: nothing leaves your machine unless you configure something that sends it, and this page lists every one of those somethings.
Last updated 11 August 2026 · Kontour AI LLC
Station is self-hosted. Your conversations, your files, and your knowledge base live in Station's own storage on your machine. There is no Kontour-operated server in the middle, no account required to use it, and no identifier derived from an account.
Everything in the next section describes a case where data does leave — and every one of them requires you or your operator to configure it first.
App version, operating system, and CPU architecture at startup. It is off unless three things are all true: an ingestion endpoint is configured, telemetry is enabled, and you have seen and acknowledged the in-app disclosure of exactly what is sent. The identifier attached to it is a random value generated once on your install and hashed before sending — it is never derived from an account, and Station never reads another vendor's credential files to build one.
Performance metrics and traces, sent only when an operator configures an OpenTelemetry endpoint — to that endpoint, which is theirs, not ours. The identifier on these is a random value created once for your install and stored alongside Station's own data. It is stable, so an operator can group one install's metrics together, and it is not derived from your machine name, your username, or any account.
Until 11 August 2026 that identifier was derived from your machine's hostname and operating-system username, and this page said so. It is recorded here rather than quietly replaced, because a privacy policy that silently improves is one you cannot check.
Documents and text you add to Knowledge are stored and indexed locally. If you configure a remote embedding provider, the document chunks and your search queries are sent to that provider so it can produce embeddings. Which provider that is, is your choice.
Microphone audio is captured only while you have explicitly started a voice session. It travels over your authenticated connection to your own Station, which forwards it to the speech provider you configured. For the built-in provider, that is Amazon Bedrock. If you use a browser's own speech features instead, that browser's handling and retention are the platform's, and we cannot describe them for you.
Used for one thing: scanning a pairing code to connect a device to your Station. Frames are decoded on the device, the pairing value is used to pair, and no imagery is sent to us or to a third party.
None of it. Station does not track you across apps or websites, does not contain an advertising SDK, and there is nothing here to sell — the data that would be worth selling never reaches us in the first place.
Station keeps a single inventory of what it collects. The app's iOS privacy manifest, its Play Data Safety answers, and this page are all derived from that one inventory, and a test in the app's build fails if any of them drifts from it — including if a new event starts collecting something the inventory does not cover.
This page was generated from inventory 788a17b779346da5.
One honest limitation: the site policy at /privacy/ links each statement to the source that implements it. This page cannot, because the Station app's source is not public yet. When it is, the links belong here too.
Station's data is on your machine, so deleting it is yours to do — remove the app, or clear its home directory, and it is gone. If you configured an endpoint that received telemetry or diagnostics, that store is yours or your operator's to clear.
Anything else — questions, a deletion request for something we hold, or a correction to this page: hello@kontourai.io.