Start a Discovery
Browse documentation

Connecting Databricks

Share tables to Neuralift over Delta Share, and receive segment-labeled results back the same way — no credentials in either direction.

Databricks connects to Neuralift through Delta Share: you create a share in your own workspace, add Neuralift as a recipient, and Neuralift reads the tables you chose to include. Nothing is copied until you sync it, and no usernames, passwords, or tokens are stored: the share is the connection. When results are ready to come back, the same mechanism runs in reverse — Neuralift shares each published run to your metastore, covered in Receiving results in Databricks below.

How this connection works

The link is built on Delta Sharing, the open protocol for secure data sharing developed by Databricks (branded OpenSharing in current Databricks documentation; see Databricks’ sharing overview). Sharing happens at the governance layer of your lakehouse: you stay in control of exactly which tables are shared, access is revocable on your side at any time, and Neuralift never receives workspace credentials. Both directions are Databricks-to-Databricks shares, authenticated by metastore identity alone.

Before you begin

  • You need the admin role in your Neuralift organization; see Roles & permissions.
  • In your Databricks workspace you need a metastore admin, or a user with the CREATE SHARE and CREATE RECIPIENT privileges (or a colleague who has them), and Delta Sharing enabled on your metastore.
  • Decide which tables you want Neuralift to be able to read. See Data requirements for what a good input table looks like. And remember: Neuralift does not accept any data with PII.

Copy your Sharing ID from Neuralift

  1. In Neuralift, go to Settings → Data Connections and click Add connection.
  2. Under Warehouse type, select Databricks.
  3. In the instructions box, click Sharing ID to copy Neuralift’s sharing identifier to your clipboard. Keep this dialog open; you’ll come back to it.

Create the share in Databricks

  1. In your Databricks workspace, create a Delta Share and add to it the tables you want Neuralift to read.
  2. Add Neuralift as a recipient of that share, using the Sharing ID you copied as the recipient’s sharing identifier.
  3. Note the provider name and share name you used; you’ll enter both in Neuralift.

The same three steps in SQL, if you prefer it to the Catalog Explorer UI:

CREATE SHARE neuralift_share;
ALTER SHARE neuralift_share ADD TABLE my_catalog.my_schema.customers;

CREATE RECIPIENT neuralift USING ID '<sharing-id>';
GRANT SELECT ON SHARE neuralift_share TO RECIPIENT neuralift;

Finish the connection in Neuralift

  1. Back in the Add data connection dialog, enter a Connection name (lowercase letters, digits, and dashes; unique within your organization).
  2. Enter the Provider name: the provider identifier for your Databricks account, for example <your-org-name>.
  3. Enter the Share name: the name of the share you created, for example <neuralift-share>.
  4. Click Save & connect.

The dialog closes and a Setting up … connection… notice appears. After a moment the new connection card appears with an Active badge (or Error with an explanation) and gains Browse tables and Test buttons.

Note. Accepting your workspace as a data provider is a one-time step Neuralift completes on its side. If the card lands on Error directing you to contact support, that acceptance is still in progress — your Neuralift team will confirm once the connection is live.

Receiving results in Databricks

Delivery runs over the same Delta Sharing link, in the opposite direction: Neuralift provisions a share for your organization and adds your metastore as a Databricks-to-Databricks recipient. There are no tokens or activation URLs — the recipient is bound to your metastore identity. Configuring delivery is a once-per-organization admin task; the delivery workflow around it is covered in Delivering results to your warehouse.

On your side:

  1. Find your Databricks metastore ID in your workspace under Unity Catalog settings, and enter it when configuring delivery under Settings → Data Connections → Share your results.
  2. Once the panel shows Active, the share appears in your workspace under Delta Sharing → Shared with me, with the share and provider names shown on the panel.
  3. Have a metastore admin mount the share as a catalog and grant access to your users:
CREATE CATALOG neuralift_results USING SHARE `<provider-name>`.`<share-name>`;
GRANT USE CATALOG, USE SCHEMA, SELECT ON CATALOG neuralift_results TO `data-analysts`;
  1. Each published run your Neuralift team delivers appears as a table named outputs.sld_<run_id> in the mounted catalog. New runs appear automatically; there’s nothing to re-mount.

Two things to know:

  • The recipient metastore ID is fixed once provisioned. Moving delivery to a different metastore means using Reset & re-provision on the delivery panel in Neuralift, then mounting the new share.
  • Cross-cloud and cross-region sharing just works. Delta Sharing handles it at the protocol level; there is no networking to set up.

Troubleshooting

  • The connection lands on Error. The card shows an explanation. Most often the Provider name or Share name doesn’t match what you created in Databricks, or Neuralift wasn’t added as a recipient with the exact Sharing ID from the dialog. Fix the share in Databricks, then remove and re-add the connection.
  • The Error card says to contact support. Neuralift’s one-time acceptance of your workspace as a provider hasn’t completed yet. Nothing is wrong on your side; your Neuralift team will confirm when the connection is live.
  • The connection is Active but tables are missing. Only tables added to the share are visible. Add the missing tables to the share in Databricks, then use Sync tables in the Source tables sheet to re-discover them.
  • Databricks isn’t offered as a warehouse type. It isn’t enabled for your organization; contact support@neuralift.ai.

Next steps