Databricks announced the public preview of Lakeflow Designer, a visual, no-code interface for data preparation built directly on the Databricks platform, according to a post by Jason Messer, Emanuel Zgraggen, V Maharajh, Matt Jones, and Tracy Yang.
Lakeflow Designer was first introduced at Databricks’ Data and AI Summit last year. The public preview follows a period of early customer work that Databricks describes as refining the product and identifying where it is most useful.
How Lakeflow Designer works
Lakeflow Designer represents each transformation step as a visual operator on a drag-and-drop canvas, with data previews available at each stage. Users can either build workflows manually or describe what they want in plain English and have Genie Code, Databricks’ agentic coding assistant, generate or modify the workflow directly.
Databricks states that Genie Code in this context has access to more than column names — it can use Unity Catalog metadata, table descriptions, lineage, popularity, and example queries to identify the right assets for a task. The post describes the system as capable of iterating: “if a join fails or returns no rows, Genie Code can evaluate the result and try an alternative approach.”
Every transformation in Lakeflow Designer generates production-ready Python code, which Databricks says can be reviewed, versioned in Git, and integrated into larger production workflows. Workflows can be scheduled and operationalised through Lakeflow Jobs.
Differentiation from existing low-code tools
Databricks identifies two points it says distinguish Lakeflow Designer from existing self-service data preparation products.
First, it runs natively on Databricks with no data movement to a separate tool. Data remains governed by Unity Catalog, and Databricks frames this as simplifying the overall data stack by not requiring separate licensing, permissions, and administration for a standalone low-code product.
Second, there is no per-user licence fee. Databricks states: “You only pay for the compute you use.” The post describes per-seat pricing in existing tools as a primary adoption barrier, forcing teams to decide upfront which users receive access.
Customer feedback
The announcement includes statements from customers. Phelipe Naman, Data and Analytics Architecture Tech Lead at Sabesp, said in Databricks’s announcement: “Lakeflow Designer expands autonomy for business teams, enabling the efficient creation of data views through natural language and best practices, while ensuring data consistency, governance, and reliability.”
Mark Wallington, Audit Data and AI Partner at KPMG UK, said in Databricks’s announcement: “Equipping our practitioners with Lakeflow Designer enables a visual, low-code and AI assisted workflow that scales and democratises our ability to translate complex and varied data sets into meaningful insights.”
Matheus Polycaropo, Data Engineering Leader at Serasa Experian, said in Databricks’s announcement: “Lakeflow Designer is a key enabler for scaling data engineering beyond the core technical team on Databricks.”
Carlos Gumz, Data Lead at Hering, said in Databricks’s announcement: “With the adoption of Lakeflow Designer, we simplified the construction of data pipelines and elevated the quality of analyses through low-code development and AI capabilities powered by natural language.”
Availability
Lakeflow Designer is described as currently available in all Databricks workspaces. To access it, users click the “New” button in the workspace and select “Visual data prep.” Databricks notes the feature may need to be enabled by an admin in the preview portal if the option is not visible.
The post does not provide information on what usage charges apply for the compute consumed by Lakeflow Designer workflows, or when the tool is expected to move from preview to general availability.