Databricks renamed products: every old name and its new name

Databricks has renamed most of its products since mid-2025. DLT is now Lakeflow Spark Declarative Pipelines, Databricks Jobs is Lakeflow Jobs, Asset Bundles are Declarative Automation Bundles, the Databricks Assistant is Genie Code, Vector Search is AI Search, Delta Sharing is OpenSharing and Genie spaces are Genie Agents. Almost none of it breaks code: the old names keep running, and the few things you should change are the ODBC driver and a handful of Python names.

Every rename, newest first

The dates are when the rename appeared in Databricks' release notes. Old names stay in older docs, blog posts and runbooks long after a rename, so expect to meet both for a while.

Check your code for old names

Paste a notebook, a job config or a runbook. Each old name shows with its new one, and whether you need to act:

Try itFind the old Databricks names in your code

Sample

Pipeline codeDriver and jobsSearch and runbook

Code or text to checkimport dlt from pyspark.sql.functions import col @dlt.view def raw_orders(): return spark.readStream.table("shop.bronze.orders") @dlt.table def orders_clean(): return spark.readStream.table("raw_orders").where(col("amount") > 0) dlt.create_streaming_table("customers") dlt.apply_changes( target="customers", source="customers_cdc", keys=["id"], sequence_by=col("updated_at"), )

1. line 1import dlt→becomesfrom pyspark import pipelines as dpStill runs

   dlt still imports; Databricks recommends the pyspark.pipelines module.

2. line 4@dlt.view→becomes@dp.temporary_viewStill runs

   Same behaviour under the new name.

3. line 8@dlt.table→becomes@dp.table or @dp.materialized_viewStill runs

   @dp.table now makes streaming tables; a batch table becomes @dp.materialized_view.

4. line 12dlt.<function>→becomesdp.<function>Still runs

   Once the import is dp, every dlt. call becomes dp.

5. line 13apply_changes()→becomescreate_auto_cdc_flow()Still runs

   Same signature; the new name is also the one Apache Spark 4.2 uses.

What actually changes in code

Most renames are words on a screen. These 5 reach your files.

Pipelines. import dlt still works, but the module Databricks recommends is the same one Apache Spark's own declarative pipelines use, so the code runs on open-source Spark too:

# Before
import dlt

@dlt.table
def orders_clean(): ...

# After
from pyspark import pipelines as dp

@dp.materialized_view   # a batch table; @dp.table now means a streaming table
def orders_clean(): ...

@dlt.view becomes @dp.temporary_view. A few dlt names stay on purpose: the classic SKUs on your bill still start with DLT, and the event log schemas with dlt in their names have not changed.

Change data capture. apply_changes() is now create_auto_cdc_flow() and APPLY CHANGES INTO is AUTO CDC INTO, with the same arguments. The old names still run. The new Python name is also the one Apache Spark 4.2 adopted, covered in Auto CDC in Spark 4.2 and the change data captureChange Data CaptureCopying only the rows that were added, changed or deleted in a source database, read from its change log, instead of copying the whole table again. concept.

The ODBC driver. This is the one rename with real work in it. Databricks no longer ships new Simba versions, and the ones you have are supported for 2 years from February 2026. The new driver changes 3 things and nothing else:

Driver={Simba Spark ODBC Driver};Host=...     ->  Driver={Databricks ODBC Driver};Host=...
SIMBASPARKINI                                 ->  DATABRICKSODBCINI
/opt/simba/spark                              ->  /opt/databricks/databricksodbc

Both drivers can be installed side by side, so you can move one application at a time.

AI Search. The Python package is now databricks-ai-search and the client AISearchClient. databricks-vectorsearch forwards to the new package with a deprecation warning, and VectorSearchClient is an alias of the same class, so nothing breaks. The REST API paths still say vector-search. The rename came with a real feature: full-text indexes that need no embeddings at all, alongside vector searchVector SearchFinding the stored items whose embeddings are closest to a query's embedding, so a search matches meaning rather than exact words..

System tables. Queries on system.workflow should move to system.lakeflow, which holds the same job tables plus newer ones such as pipelines.

The Genie names, untangled

Genie had the busiest year. 3 products now share the name:

• Genie One is the chat screen for business users. It started as Databricks One, became Genie in April 2026 and Genie One in June. Genie One covers setup and cost.
• Genie Agents are the curated question rooms a data team builds over chosen tables, called Genie spaces until July 2026.
• Genie Code is the coding agent for engineers, the Databricks Assistant until March 2026. Genie Code pricing covers its free allowance.

Inside Genie Agents, Research Agent became Agent mode, and the tuning features were renamed for what they do: value sampling is prompt matching, example values are format assistance and value dictionaries are entity matching. In metric view screens, Fields replaced Dimensions. Genie vs Genie One vs Genie Code puts the 3 side by side.

Why the names keep changing

The renames follow 3 lines, each stated in Databricks' own notes:

1. Lakeflow is the umbrella for data engineering. Ingestion (Lakeflow Connect), pipelines and jobs now sit under one name, and the Pipelines and Workflows menu items merged into Jobs & Pipelines.
2. Open-source names match. Spark Declarative Pipelines is the name Apache Spark uses from 4.1, and OpenSharing followed the release of an open-source OpenSharing project.
3. Names describe the product. Databricks said "assets" in Asset Bundles confused people because the word means several things on Databricks, and Vector Search became AI Search when it started doing keyword search without vectors.

What did not change: Delta Lake is still Delta Lake (only the sharing product was renamed), Unity Catalog is still Unity Catalog, and the databricks bundle command is the same.

If you are building pipelines with the new names from scratch, our data engineering course part on running the medallion pipeline on Databricks uses them throughout.