Genie now answers across spaces, reads your Google Drive, and lives on your phone. The work that makes those answers trustworthy is still ours.
Update, June 2026: Genie is moving to pay-as-you-go pricing on July 6, 2026, across Genie Code, Genie Spaces, and Genie. Each user gets 150 free DBUs per month, then usage is billed per DBU, with the underlying compute billed separately. Databricks also added usage budgets so you can set a ceiling before adoption takes off. Read the full breakdown: Genie Goes Pay-As-You-Go: Pricing and Budgets Explained
A Monday morning, somewhere inside a large company.
A regional director opens Databricks on her phone. She types one short question. "Why did the South region miss its Q1 revenue plan?"
A year ago, that question would have meant a chain of messages, three different dashboards, a SharePoint policy doc, and probably a Slack thread with a data engineer. Today, one chat answers it. Genie pulls from a certified sales Genie Space, joins it with the supply chain Space, reads the regional pricing memo from Google Drive, and writes a short answer with sources.
This is the world Databricks just announced.
At the Data + AI Summit in June 2026, Databricks announced the next generation of Genie. The summary, in plain English:
mycompany.databricks.com and unified login.A new agent architecture sits underneath all of this, with stronger reasoning models that can synthesize across multiple data domains.
Two shifts are happening at the same time, and they matter more than the new chat box.
The first shift: the unit of trust moves from a single Genie Space to your whole governed catalog. Until now, a Genie Space was a small, careful boundary. A team curated a few tables, added verified metric definitions, and that was the trust zone. The new Genie still relies on that work, but it walks across many of those zones in a single conversation. That means the quality of your overall Unity Catalog, not just one Space, is now the quality ceiling for every answer.
The second shift: unstructured context becomes a first-class citizen. A pricing memo on Google Drive, a SharePoint policy page, a runbook in Confluence. These have always been part of how the business actually thinks. Now they are part of what Genie reads, alongside your Delta tables and the new FILE type in Unity Catalog. That is a real change in what "data" means.
Here is the honest version.
Genie did not get smarter than your data. It got better at exposing how clean, well-named, well-governed, and well-documented your data is.
Genie reveals how good your engineering is.
If your Silver tables have meaningful column comments, sensible names, owners, and tags, Genie will look brilliant on top of them. If they do not, Genie will look confused, and your team will assume the AI is wrong. Most of the time it will not be.
This is good news. The work that makes Genie trustworthy is the same work that makes a normal pipeline trustworthy. Nothing exotic.
Calm, practical, no rush.
1. Tighten Unity Catalog. Add column comments, table descriptions, owners, and tags. Mark your trusted tables as certified and use the new Glossary and Domains features. If you have not spent time here in a while, this is the highest-leverage hour you can put in. Start with our lesson on Unity Catalog.
2. Make your medallion layers honest. Bronze should be raw and traceable. Silver should be clean and trustworthy. Gold should be the metric layer your business reads. The cleaner this separation, the better Genie reasons across it. Walk through Medallion Architecture and pair it with Data Quality.
3. Write SQL that explains itself. Aliases that read like English. Comments on the non-obvious joins. Verified metric views with clear names. When Genie reuses your logic, it inherits your clarity or your mess. Refresh on Spark SQL and Joins and Aggregations.
4. Treat documents as data. Now that Genie can read SharePoint and Google Drive through the Unity AI Gateway, somebody has to decide which folders are in scope and which are absolutely not. That is a governance conversation, not a tooling one. Bring it up before someone else does.
Be honest about the limits. Account-level Genie, the new mobile apps, and the enterprise connectors are paid-workspace features. You will not run them on Free Edition.
You can practice everything underneath them today.
Open Databricks Free Edition, follow Start Here, and build one small medallion pipeline end to end. Add column comments. Add a verified metric view. Notice how much clearer your own queries feel after an hour of that. That clarity is the same clarity Genie will read.
If you want a guided sequence, our Workflows lesson and Debugging and Monitoring lesson cover the orchestration and feedback loop that real pipelines need.
The Data Engineer Associate exam does not quiz you on Genie features. But the May 2026 guide leans heavily on Unity Catalog governance, ABAC policies, Lakeflow orchestration, and Lakeflow Declarative Pipelines (formerly Delta Live Tables). These are exactly the foundations the new Genie sits on.
In other words: studying for the exam is studying for a Genie-ready workspace.
If you are preparing, our Databricks Certification Guide lays out a calm 6-week plan, and the calm walk through the May 2026 exam guide covers what changed.
When you are ready to test where you stand, the free DE Associate practice exam is built around the same six domains.
Pick one Silver table you own. Open it in Free Edition. Write three honest column comments and a one-sentence table description. Then close the laptop.
Tomorrow, imagine someone asking your AI a question that depends on those columns. Notice how much more confidently you can predict the answer.
That is the work. Genie just made it visible.