Databricks launches Genie Code: agentic engineering for data work

Databricks’ Genie Code is an AI agent that writes, tests, and executes data engineering code within the Databricks Lakehouse platform. The agent targets a persistent bottleneck in product organizations: the gap between a business question and the data pipeline needed to answer it.

Rather than generating code snippets for a human to review and run (the copilot model), Genie Code operates as an autonomous agent that can build entire data workflows — creating tables, writing transformations, running tests, and iterating until the output meets quality checks. Human data engineers review the completed work rather than supervising each step.

For product managers, this matters because data availability is one of the most common blockers for product decisions. If AI agents can reduce the time from “I need this data” to “here is the data” from days to hours, PMs gain faster feedback loops for product experiments, A/B test analysis, and usage pattern investigation.

The broader pattern: AI agents are not just assisting knowledge workers but taking over the execution of well-defined technical tasks, with humans shifting to specification and review roles.