Connectors

A connector pulls data from an external source into a dataset, so the data your models learn from stays fresh without anyone uploading files by hand.

Supported sources

Every sync lands as a new dataset version — the same validated, versioned asset as a manual upload. Models keep training against pinned versions; a sync never changes data underneath anything.

Credentials are handled carefully

Connector credentials are stored encrypted and are write-only: once saved, they are never shown back in the UI or the API, and they never appear in audit records. When editing a connector, leaving the secret field blank means "keep the stored secret" — you only type it again to change it.

For centralized management (one database password shared by several connectors, rotated in one place), store the credential in Secrets and pick it from the secret picker in the connector form instead of typing it inline.

Working with connectors in the Studio

1. Go to Connectors in the sidebar and create one: choose the type, fill in the connection details, and provide credentials (inline or from Secrets).
2. Optionally set a sync schedule so the connector sweeps on its own.
3. Click Sync now to run immediately. A sync is asynchronous — the row shows it running, then flips to its result.
4. Open the connector's detail page for the sync history: every run, when it happened, whether it succeeded, and a link to the dataset version it produced.

Reading the statuses

• A disabled connector is not broken — it just runs no scheduled sweeps. You can still trigger a manual sync.
• A failed sync is shown in red on that run, with the error message; the connector itself is fine, and the datasets it produced earlier are untouched.

Next

• The data has arrived — shape it: Feature sets.
• Centralize the passwords: Secrets.