Support
Documentation covers capabilities, tips, skills, MCP, instructions and the data science, engineering and dashboard agents. A demo video is also linked.
Learn moreAgentic AI that builds and maintains data work inside Databricks

Databricks Genie Code is a platform-native AI agent that plans and runs data science, ML, pipeline and dashboard tasks inside the Databricks workspace, grounded in Unity Catalog governance and metadata.
Databricks Genie Code is an AI agent aimed at data teams that already work inside the Databricks workspace. Rather than acting as a general coding chatbot, it handles multistep tasks across data science, machine learning, data engineering and business intelligence, and it keeps context as work moves from one task to the next. The agent is grounded in Unity Catalog metadata, semantics and governance. In practice, that means it can pick authoritative tables, follow dependencies between data and AI assets, and stay inside the permission model the organization has already set up. Users can also point it at specific tables, notebooks, files, folders and dashboards, attach screenshots or diagrams as context, and describe a complex job in plain language. Genie Code then drafts a structured plan that the user reviews and approves before anything runs. Coverage is broad. For data science it automates exploratory analysis, from locating and cleaning data to producing shareable reports. For machine learning it engineers features, trains and evaluates models, deploys them, and adjusts endpoint configuration based on observed traffic. For pipelines it builds ETL workloads and Spark Declarative Pipelines through conversation, optimizes queries, and monitors and fixes issues in production. For BI it plans datasets, defines metrics and generates dashboards. Teams can extend behavior with Agent Skills that package reusable code and best practices, persistent custom instructions, and MCP connections that expose outside tools and data. Genie Code sits at the platform-native end of the market. General assistants such as editor-based coding tools may be more flexible across stacks, but they lack direct awareness of a lakehouse catalog. Genie Code trades that flexibility for depth inside Databricks, which makes it most relevant for organizations that have standardized on the platform and want an agent that respects existing governance. Customer examples on the site include large enterprises in energy, media, pharmaceuticals and retail.




Documentation covers capabilities, tips, skills, MCP, instructions and the data science, engineering and dashboard agents. A demo video is also linked.
Learn moreWorks inside the Databricks workspace with Unity Catalog, Notebooks, Lakeflow and AI/BI Dashboards. MCP support lets it retrieve context from external tools, data and workflows.
Learn moreAutonomous exploratory analysis, ML feature engineering, training and deployment, ETL and Spark Declarative Pipeline building, dashboard generation, agent plans with approval, Agent Skills, MCP support, custom instructions, data discovery, image uploads and assets as context.
Learn more




