New in Signals: Databricks support and warehouse sync controls

Signals can now connect to a Databricks warehouse, alongside Snowflake and BigQuery. We've also added a sync frequency setting for warehouse attribute groups.

Databricks is now supported

Databricks joins Snowflake and BigQuery as a supported warehouse for Signals. This covers both places Signals reads from a warehouse: backfilling a stream attribute group from your atomic events table, and syncing pre-calculated attributes from your own tables.

You can still deploy Signals without a warehouse connection if you only need attributes calculated from the live event stream. See Set up Signals for how to connect one.

Choose how often warehouse attributes sync

Warehouse attribute groups sync hourly by default. If your source table updates less often than that, hourly syncing costs you warehouse compute for no new data.

You can now pick a sync frequency of 6, 12, or 24 hours when you create the group in Console. In the Python SDK, set refresh_rate on your ExternalBatchAttributeGroup:

python

from datetime import datetime, timedelta, timezone

attribute_group = ExternalBatchAttributeGroup(
    name="user_transactions",
    version=1,
    attribute_key=domain_userid,
    batch_source=data_source,
    backfill_since_tstamp=datetime(2026, 6, 1, tzinfo=timezone.utc),
    refresh_rate=timedelta(hours=6),
    fields=[...],
)

For the full configuration reference, see Sync warehouse tables to Signals.