Databricks already gives data teams a place to store, govern, and work with company data. The harder part is making that data useful across the business, especially when an answer depends on company terminology or information that still lives in operational tools outside the warehouse.
Supper connects directly to Databricks so employees can ask questions, investigate the results, and work with data from other connected sources. Databricks remains responsible for the warehouse, compute, and Unity Catalog governance. Supper uses that environment alongside the company's business definitions and other data sources to answer questions in context.
For Databricks customers, getting started is fairly light. After the company provisions a system account, Supper can inspect the permitted data, build an initial semantic model, and create an evaluation corpus for testing. Users can typically begin asking questions within one or two days.
The roles of Supper and Databricks
Companies use Databricks for data storage, processing, engineering, machine learning, and governance through Unity Catalog. Its AI/BI products also include dashboards and Genie Agents, which let users analyze governed Databricks data through natural language.
Supper gives a wider group of employees a way to work with that data. Users can ask questions conversationally, continue an analysis through follow-up questions, create charts, and save recurring work as reusable Skills. Supper can use SQL to query the warehouse and Python when an analysis requires calculations or additional processing.
Many business questions also depend on context that is difficult to capture in the database alone. The meaning of a metric may depend on an internal definition, an exception to the normal calculation, or a relationship between systems. Supper maintains that context in its semantic model and applies it when answering questions.
Connecting Supper to Databricks
A company begins by provisioning a Databricks system account for Supper. Supper uses that account to inspect the database structure available to it, create the initial semantic model, and generate an evaluation corpus that tests how well the model answers questions.
The customer's part of the initial connection generally takes about five minutes. Supper completes the automated setup and testing in the background, with users typically able to begin asking questions within one or two days. The exact timeline depends on the size and condition of the Databricks environment.
Supper queries Databricks directly. The company can keep its existing warehouse architecture, and Databricks continues to process the warehouse queries.
The semantic model develops further after launch. Questions from real users expose missing definitions and areas where Supper needs more context. Supper can identify many of these issues automatically, while decisions involving company-specific business rules remain with the customer. During the first month, this review usually requires anywhere from a few minutes to a few hours of customer involvement.
How Supper uses Unity Catalog
Unity Catalog gives Supper an established record of Databricks assets and access rights. Supper can import this information along with semantic metadata that the company already maintains in Databricks.
Companies can decide how Supper should treat the existing model. If the Unity Catalog model is complete and trusted, Supper can accept it as the approved definition of the data. If it contains useful information but needs more context, Supper can use it as a starting point and supplement it.
Supper's semantic model can include company terminology, calculation instructions, known data issues, table relationships, and reusable analytical processes. This knowledge is maintained at the model level so it can be applied consistently across users and questions.
This also saves data teams from rebuilding context they have already documented. Work completed in Unity Catalog can carry into Supper, while Supper can maintain the additional context needed for business-facing analysis.
Analysis across Databricks and SaaS data
Important business information often remains spread across operational applications. A company may keep historical revenue data in Databricks while current sales activity remains in its CRM.
Take renewal risk as an example. Answering the question may require product usage and billing history from Databricks, along with current account activity from the CRM. Supper can determine which sources are relevant, query them according to the user's permissions, and apply the company's definition of an at-risk account.
The person asking the question does not have to know where each field lives or how the systems relate. They can ask the business question directly and continue investigating the result within the same conversation.
This lets teams analyze Databricks alongside connected finance, sales, and operational systems. It also means that a useful answer does not have to wait until every relevant field has been copied into the warehouse and modeled there.
Permissions and administrative oversight
Supper can import Unity Catalog entitlements and incorporate them into its own permissions system. Access rules can also cover the SaaS sources connected to Supper, giving administrators control over which information is available to each person.
Permissions are evaluated for the individual user. Giving Supper a system account does not give every employee unrestricted access to the data available through that account.
Administrators can review how people are using company data through Supper's Admin Control Center. They can see the questions being asked, investigate how data is being accessed, and review issues found in the semantic model. This gives the data team continued oversight after AI analytics has been made available to the wider company.
When should a Databricks customer use Supper?
Databricks AI/BI may cover the use case when analysis remains inside Databricks and the intended users already work comfortably in its environment.
Supper is relevant when a company wants to make Databricks data accessible to a wider business audience or when questions regularly depend on information outside the warehouse. It also gives the data team a dedicated place to manage company-specific definitions and review how employees use AI to access data.
Supper uses seat-based pricing. The cost is tied to the number of users rather than the number of tokens consumed during an analysis, making costs easier to anticipate when questions require several analytical steps.
Frequently asked questions
Does Supper replace Databricks?
No. Databricks continues to provide the warehouse, compute environment, and Unity Catalog governance. Supper connects to Databricks and gives business users another way to analyze the data stored there.
How does Supper connect to Databricks?
The customer provisions a Databricks system account for Supper. Supper uses that connection to inspect the permitted schema, build its initial semantic model, run evaluations, and query Databricks directly.
Does Supper copy Databricks data into another warehouse?
Supper queries Databricks directly for warehouse analysis. Companies do not need to migrate or duplicate their Databricks warehouse before they can begin using Supper.
How is Supper different from Databricks Genie?
Databricks Genie Agents provide natural-language analysis within the Databricks AI/BI environment. Supper provides a business-facing analytics experience with its own semantic model, reusable Skills, administrative controls, and support for questions that use Databricks alongside connected SaaS sources.
Can Supper use an existing Unity Catalog model?
Yes. Supper can treat Unity Catalog as the approved model or use it as an input that can be supplemented with additional business context. The company chooses the approach that matches its existing Databricks model.
How long does the Databricks setup take?
Provisioning the connection generally takes about five minutes of customer work. Supper can usually prepare the initial semantic model and evaluation corpus within one or two days, after which users can begin asking questions.