Salesforce links with Databricks, Snowflake to build stronger enterprise data foundations

Salesforce is kicking off Dreamforce with some major partner announcements.

Today, the Marc Benioff-led customer relationship management (CRM) software leader announced that its proprietary Salesforce data Cloud, which brings together data points from different sources to host unified customer profiles in real-time, will support bi-directional data sharing and access with Databricks’ data lakehouse platform. 

The move, set to go live at a later stage, will allow joint customers of the companies to enrich their datasets and power additional use cases, including building and deploying more capable models targeting different business-critical problems.

“Access to trusted, governed data is critical for every company, and the ability to combine that data with AI is now essential to remain competitive. Delivering best-in-class integrations with Salesforce builds on our longtime partnership…, unlocking massive value for our mutual customers, Adam Conway, SVP of Products at Databricks, said in a statement.

However, as it turns out, Databricks isn’t the only one linking up with Salesforce for these benefits. Just last Friday, Snowflake, the Ali Ghodsi-led company’s arch-rival, also announced a similar strategic partnership to connect its data cloud with Salesforce’s data platform.

How does integration with Salesforce data cloud work?

Traditionally, moving data across platforms involved complex extract, transform and load (ETL) processes, forcing teams to invest significant resources in data engineering or ETL tools.

With the new zero-ETL integration, dubbed BYOL or Bring Your Own Lake, Salesforce will make it as if the unified data (from its own and the other connecting platform) is housed at a single location – while maintaining the highest levels of governance, security and trust at the same time.

At the core, Salesforce explained, BYOL features two key capabilities: Data Sharing and Data Federation. The former allows users to access Salesforce data cloud information from within the connecting platform, be it Databricks or Snowflake, while the latter allows the information from these two platforms to be accessed from within the Salesforce data cloud.

This bi-directional data movement has two-way benefits for joint users. On Salesforce, external data from Snowflake or Databricks can help teams build richer customer profiles and use them across existing applications and processes across Customer 360 to deliver better customer experiences. 

Meanwhile, Salesforce’s CRM data on Databricks and Snowflake will enhance the data foundation of these platforms, allowing teams to power downstream use cases, including AI/ML or deeper analysis, for better business outcomes.

For instance, by connecting Salesforce CRM data like transaction history and credit score with market analysis and economic trends, a Databricks user can create custom cross-selling models that recommend additional products or services to advisors based on real-time engagement data from clients.

Similarly, combining Salesforce data such as sales and website visits with external market trends in Snowflake can help retailers identify changes in customer behavior, allowing for smarter inventory management and marketing decisions.

Databricks models can be moved to Salesforce

While Snowflake users will also be able to create custom AI models with the enriched data, the announcements suggest that Databricks users will be the first ones to get support to move their models to the Salesforce data cloud to power any application on the platform.

“Data scientists and developers want to use their preferred tools and ML frameworks to build models that can drive predictions across the Salesforce platform. Databricks unifies the data and AI platform so those AI/ML teams can now build, train and govern their AI models in Databricks and easily bring those models into Salesforce and apply them across the Customer 360 platform,” Conway told VentureBeat.

This movement of models will be executed through Salesforce’s Bring Your Own Model experience delivered through Einstein Studio announced last month, the company confirmed. 

“Customers (will) have the ability to view and control access, sharing capabilities, and permissions at the admin level through Unity Catalog. These permissions will apply across platforms so mutual customers will control the flow of their data/models as they move through the Lakehouse and Salesforce,” Conway noted while detailing the company’s effort on the security front.

When can teams unify their data?

Even though BYOL integrations have been announced for both Databricks and Snowflake, the connectors are not fully ready yet. For Snowflake, Salesforce confirmed that only BYOL Data Sharing is generally available and Data Federation is expected to roll out at a later stage.

For Databricks, on the other hand, the integrations are still being developed.

“Our engineering teams are working on building these integrations now and we expect them to be available for customers to preview early next year. We have thousands of mutual customers and it’s important that we make it as easy as possible to access and securely share data across platforms without having to maintain complex or costly data pipelines in multiple places,” Conway said.