Genie One and Genie Code: what they mean for data teams

At Data + AI Summit 2026, Databricks made Genie One generally available. It is an assistant that lets business users ask questions of their data in plain language. Alongside it came Genie Code, an AI coding agent for data engineering and machine learning.

Together they change who works with data, and how.

What changes for business users

Business users no longer need to wait for a new dashboard. They can ask Genie directly, for example: "Which regions missed target last quarter?" Genie also connects to many enterprise apps, and it can be used inside Microsoft Teams.

This is good news for analysts, who spend less time on one-off requests. It also raises the stakes. If the underlying tables are messy, or two teams define revenue differently, Genie will give confident answers that do not match.

What changes for engineers

Genie Code writes, explains and fixes pipeline and ML code. It speeds up routine work. But code that is written faster still needs review, tests and clear ownership. Teams that skip this step trade speed today for incidents tomorrow.

How we roll out Genie

• Start with one team and ten questions. Pick the questions that team asks every week, and make sure Genie answers them correctly.
• Fix the foundations first. Clean gold tables, clear column descriptions and governed metrics in Unity Catalog do more for answer quality than any prompt.
• Use permissions you already have. Genie respects Unity Catalog access, so people only see data they are allowed to see.
• Measure trust. Track how often answers are accepted, corrected or escalated, and improve the space each week.

Where to begin

If you already run Databricks, a first Genie space can usually be set up quickly. The harder and more valuable work is curating it. That is where a small, focused team pays off.