Will AI replace traditional BI dashboards?

Summary

• No. On Databricks, chat and dashboards are two views of the same governed data. Genie One lets business users view AI/BI dashboards, ask data questions in natural language, and use Databricks Apps from one entry point.
• Every published AI/BI dashboard includes a Genie Agent by default. Viewers get an Ask Genie button for questions about the dashboard or a selected chart, and for explanations of a change on supported chart types.
• Dashboards keep the recurring jobs. A published dashboard holds the version the author approved until you publish again, can refresh its data on a schedule, can send subscriptions by email or to chat channels, and can be embedded in external applications.
• Metric views define a KPI once. Dashboards, Genie Agents and SQL alerts can query the same metric view.
• Check what a dashboard's Genie Agent can read. It answers from the full dataset query results, including columns and rows that no chart displays.

Will AI replace traditional BI dashboards?

No. AI changes how people reach an answer, but the dashboard stays the shared reference that a team reviews on a schedule. Databricks builds the two to work together: Genie One puts AI/BI dashboards, natural-language chat and Databricks Apps behind one entry point, and every published dashboard can carry its own Genie Agent. We believe the durable pattern is both.

How do AI chat and dashboards compare?

Need	AI/BI dashboards	Genie (Ask Genie, Genie Agents, Genie One)
Starting point	An author defines datasets and visualizations. Viewers apply filters, cross-filtering and drill-through.	A user asks a question in natural language.
What comes back	The author's charts and tables.	A generated SQL query, a results table and visualizations.
Recurring review	Scheduled refreshes, plus subscriptions that send snapshots to email or chat channels.	Genie One supports scheduled chat tasks.
Why a number moved	The chart shows the change.	On supported chart types, Ask Genie explains the change and returns the top contributing factors.
Reaching beyond the workspace	Share with account users who have no workspace access, or embed in an external site or application.	Ask Genie works in basic embedding. For external users, use the Genie Conversation API.

What does Genie add to a published dashboard?

Publishing a dashboard turns on the Enable Genie setting by default. Databricks generates a Genie Agent from the dashboard's datasets and visualizations, and viewers see an Ask Genie button. They can chat with the data instead of relying only on predefined visualizations, select a chart and ask about it, or right-click a supported visualization and ask Genie to explain the change. Genie then enters Agent mode to analyze the change and identify the top drivers. Authors can turn the setting off or link an existing Genie Agent, and the generated agent updates when you republish.
AI also helps build dashboards. AI/BI dashboards support AI-assisted authoring, and Genie Code in Agent mode can plan a dashboard, create datasets, add visualizations, configure filters and lay out pages from a natural-language prompt. It can only reach the data that the user has access to.

Which jobs still belong to dashboards?

Publishing creates a snapshot of the dashboard, and the published version stays unchanged until you publish again, so authors can keep editing a draft while viewers see the approved version. Authors can also schedule data refreshes, which populate the query result cache, and subscribers receive snapshots on that schedule: a PDF by email, or an image with a link in a chat channel.
Thresholds and notifications live in a separate object. Databricks SQL alerts run a query on a schedule and notify you when a condition you define is met. An alert can query a Unity Catalog metric view, so a governed KPI can drive a notification.

How do you keep chat answers and dashboard numbers consistent?

Define the KPI once. A metric view separates measure definitions from the fields used to group and filter them, so you define the metric once and query it at runtime. Metric views can be queried from SQL editors, notebooks, dashboards, Genie Agents and alerts, and they are Unity Catalog securable objects governed with standard Unity Catalog privileges. In a Genie Agent, trusted assets (parameterized example queries and SQL functions that an author has verified) make the answer come from that verified logic.

What should you check before you leave Ask Genie on?

Data access. A dashboard's generated Genie Agent follows the dashboard's credential model. With shared data permissions, Genie uses the publisher's credentials. With individual data permissions, each viewer's own permissions apply. In both cases Genie answers from the full dataset query results, so a dataset that includes a sensitive column or row makes it available to Genie even when no chart shows it. Define datasets with only the fields and rows that viewers should reach.

FAQs

Can viewers ask questions without leaving a dashboard?

Yes. Published dashboards include Ask Genie by default, and viewers can select a chart to ask about it. Authors can turn it off or link an existing Genie Agent instead.

Does Ask Genie work in dashboards embedded for external users?

No. Ask Genie is available in basic embedding, where viewers sign in with Databricks credentials, but not in embedding for external users. For external scenarios, use the Genie Conversation API.

Do I need a dashboard before I can use Genie?

No. A Genie Agent is built on data registered to Unity Catalog, including tables, views, metric views and materialized views. An analyst curates it with datasets, example SQL queries and text instructions, and a dashboard is optional.

Can I see what Genie ran to produce an answer?

Yes. A Genie Agent responds with the generated query and its results table, and the generated queries are always read-only.