How do data engineers and data scientists use Databricks Genie to go from raw data to analysis, pipelines, and apps?

Summary

• Genie is a family of three experiences on the same Unity Catalog foundation: Genie Code for developers and technical practitioners, Genie Agents for domain-specific data environments, and Genie One for business users.
• Data engineers use Genie Code in the Lakeflow Pipelines Editor to plan, generate, run, and fix pipelines. Databricks documents pipeline development with Genie Code as Public Preview, and Genie Code asks for approval before it runs code or updates a pipeline.
• Data scientists use Genie Code in notebooks and the SQL editor for data discovery, exploratory analysis, forecasting, and machine learning, working from tables referenced with @table_name under their own permissions.
• Data teams curate Genie Agents with datasets, example SQL, instructions, and trusted assets. Unity Catalog row filters and column masks then apply per user when business people ask questions.
• Genie One shows business users the dashboards and apps shared with them, and Genie App Builder (Beta) generates Databricks Apps from a natural-language description.

How do data engineers and data scientists use Databricks Genie to go from raw data to analysis, pipelines, and apps?

They use three experiences in sequence. Engineers and scientists work with Genie Code to build pipelines and analysis. Data teams then curate trusted data, metrics, and business rules in a Genie Agent. Business users reach those answers, along with dashboards and apps, through Genie One. All three run under Unity Catalog permissions, and people review what Genie generates before it ships.

What are Genie Code, Genie Agents, and Genie One?

Experience	Who uses it	What it does
Genie Code	Developers and technical practitioners	An AI coding and data assistant that runs in notebooks, the SQL editor, the Lakeflow Pipelines Editor, AI/BI dashboards, and MLflow. It generates and runs code, builds pipelines and dashboards, and debugs errors
Genie Agents	Data teams who configure them; users who ask	Domain-specific environments where data teams configure trusted data, metrics, and business rules. Users get SQL queries, result tables, and visualizations
Genie One	Business users	A simplified interface to ask data questions in natural language, explore AI/BI dashboards, and run Databricks Apps

How do data engineers use Genie Code?

In the Lakeflow Pipelines Editor, Genie Code in Agent mode can plan and generate a pipeline end to end, or work on an existing one. With your approval it searches tables, edits SQL or Python source files, runs pipeline updates, and reads pipeline datasets. It can add datasets with quality expectations, trace how data flows into downstream tables, and diagnose errors across several files. The editor documentation lists pipeline development with Genie Code as Public Preview.
You stay in control of each step. Genie Code asks for approval before it runs code or updates a pipeline, and you choose how often. Auto-approve is the default for new users, but Databricks calls it a productivity feature, not a security boundary, and advises keeping it off with production data. Review generated code before you run it. Teams can also shape Genie Code with workspace and personal instructions, such as code conventions or preferred libraries, and with reusable skills for domain-specific tasks.

How do data scientists use Genie Code?

In notebooks and the SQL editor, Genie Code plans a solution, retrieves relevant assets, runs code, uses cell outputs to improve results, and fixes errors. Reference tables with @table_name so it uses their Unity Catalog metadata. The documentation lists data discovery, exploratory analysis, forecasting, and machine learning as use cases, with example prompts for feature engineering, training a classification model, and hyperparameter tuning. Genie Code can access only the data your permissions allow.

How do teams build trusted answers with Genie Agents?

Data analysts curate each Genie Agent with datasets registered to Unity Catalog, example SQL queries, SQL expressions for business semantics, and text instructions. Trusted assets are parameterized queries and SQL functions whose logic an author has verified, and benchmarks measure response accuracy. A Genie Agent uses only the data sources attached to it. Unity Catalog evaluates data access as each end user, so row filters and column masks apply per user, and generated queries are read-only.

How do apps reach business users?

Genie One shows each user the assets shared with them, including AI/BI dashboards, Genie Agents, and Databricks Apps, and its chat searches across dashboards, queries, and Genie Agents. To build an app, Genie App Builder (Beta) generates a Databricks App from a natural-language description inside an App Space, shows a working preview, and lets you refine and deploy it. A workspace admin must turn on Governed agentic app-building first.

FAQs

Does Genie Code write correct code the first time?

Not always. Databricks recommends that you review any code Genie Code generates before running it, because it can sometimes make mistakes.

What do business users see in Genie One?

Only the assets shared with them. They can view and interact with AI/BI dashboards, ask data questions in natural language, and use custom-built Databricks Apps, without working in notebooks, queries, or compute settings.

What data can a Genie Agent query?

Data registered to Unity Catalog, including managed, external, and foreign tables, views, metric views, and materialized views. The agent author attaches the sources, and Genie does not query tables that are not attached.

How does Genie handle sensitive data?

Data access is evaluated with each user's own Unity Catalog permissions. Row filters and column masks defined in Unity Catalog are enforced automatically, and a question about data a user cannot access returns an empty response.