Databricks Genie Code

Databricks Genie Code is an AI agent developed specifically for data teams, aimed at automating and managing complex data workflows. The tool is designed to analyze, build, and maintain data processes using agentic AI that adapts to the user's data and governance model. Genie Code is grounded in Unity Catalog metadata, enabling it to understand data semantics, governance, and dependencies across data and AI assets within the Databricks workspace.

Genie Code supports a wide range of data team activities, including data engineering, data science, machine learning, and dashboard development. It can automate exploratory data analysis, streamline data cleaning, and generate shareable reports. For machine learning workflows, it assists with feature engineering, model training, evaluation, deployment, and even endpoint configuration and performance tuning based on observed traffic. In data pipeline engineering, Genie Code can automate ETL workloads, optimize queries, and build Apache Spark declarative pipelines through natural conversation. The tool also facilitates the planning and generation of production-ready dashboards, retrieving data assets, configuring layouts, and refining metrics with guided execution.

Additional features include the ability to define domain-specific agent skills, which package best practices, reusable code, and workflows that Genie Code can execute. It supports MCP to securely expose tools, data, and workflows for context retrieval. Users can set persistent, system-level custom instructions to share preferences and authoring methods. Data discovery is enhanced by Genie Code surfacing relevant tables, metrics, and assets based on user specifications. The tool also allows users to upload images, such as screenshots or diagrams, which Genie Code uses as context to generate calculations, logic, and workflows. Furthermore, users can manually specify assets like tables, notebooks, files, folders, and dashboards to provide precise context for task execution.

Genie Code operates across the Databricks workspace, preserving context across tasks and proactively monitoring and resolving issues to keep production data and AI systems running smoothly. It is positioned as a specialized AI agent for data teams, supporting daily workflows from experimentation to production pipelines and business intelligence dashboards.