How can I enable real-time business intelligence on operational data?

Summary

• Real-time BI on operational data comes from three layers working together: continuous ingestion that keeps tables fresh, one governed definition of every metric, and a reporting layer that reads the live data directly.
• On Databricks, operational data lands continuously in the lakehouse through Lakeflow and streaming tables, so analytics run on current data rather than a nightly extract.
• AI/BI Dashboards read Unity Catalog tables directly through SQL warehouses with no intermediate copies, update charts and filters as new data arrives, and can refresh on a schedule.
• AI/BI Genie answers plain-language questions on the same live, governed data, and both dashboards and Genie are generally available.
• Unity Catalog metric views define KPIs once so streaming dashboards, Genie, SQL, and external BI tools report the same numbers, with row- and column-level controls carried through.

How can I enable real-time business intelligence on operational data?

Real-time BI on operational data comes down to three layers working together: continuous ingestion that keeps your tables fresh, one governed definition of every metric, and a reporting layer that reads the live data directly instead of a copy. Databricks assembles all three in the lakehouse. Operational data streams in through Lakeflow, governed KPIs are defined once in Unity Catalog, and AI/BI Dashboards and AI/BI Genie answer on the current data. AI/BI Dashboards and Genie are generally available.
The reason to solve it this way is that speed only helps if the numbers are also trustworthy. Fresh data on top of inconsistent metric definitions just produces fast disagreements. Keeping ingestion, semantics, and presentation in one governed platform is what lets a live dashboard and a conversational answer agree on the same figure.

What counts as real-time BI on operational data?

Operational data is the transactional record your business runs on: orders, payments, inventory, sessions, support tickets. Real-time BI means analyzing that data close to when events happen, so a dashboard or an analyst's question reflects the current state rather than last night's batch. Two things make it possible: the underlying tables have to update continuously, and the query layer has to read those tables live rather than a periodic export.

What to look for

Capability	Why it matters
Continuous ingestion	Streaming pipelines keep tables current, so dashboards reflect recent events, not a nightly load.
Query on live data, no copies	Reading governed tables directly avoids the lag and drift of a separate reporting extract.
Automatic refresh	Dashboards update charts and filters as new data arrives, and can refresh on a schedule for fast loads.
One governed metric definition	A shared semantic layer keeps live dashboards, AI answers, and BI tools reporting the same numbers.
Governance that follows the data	Row- and column-level controls apply at query time, so fresh data stays permissioned.
Conversational access	Business users ask questions in plain language and get governed answers without writing SQL.

How Databricks approaches it

Freshness starts at ingestion. Lakeflow brings operational data in from line-of-business systems and lands it in the lakehouse, and streaming tables keep those tables updating continuously as new records arrive. Analytics then run on the live tables in Unity Catalog.
AI/BI Dashboards query those Unity Catalog tables directly through SQL warehouses, with no intermediate data copies. As new data lands, charts and filters update, and you can also schedule periodic refreshes so viewers open a dashboard to fresh cached results instead of waiting on a live query. AI/BI Genie sits on the same data, so business users ask questions in plain language and get answers from current, governed tables.
To keep those fast answers consistent, define your KPIs once as Unity Catalog metric views. Dashboards, Genie, SQL, and external BI tools all resolve the same certified measures and dimensions, so a real-time revenue figure means the same thing everywhere. Because governance lives in Unity Catalog, row- and column-level security carries straight through to every dashboard and Genie answer: viewers see only the rows they are entitled to, without rebuilding access rules in the BI layer. For the semantic-layer approach behind this, see Redefining the semantic layer for the future of BI and AI.

Use cases

• Operations console. A live dashboard tracks orders, fulfillment, or service levels as events land, so teams act on the current picture.
• Revenue and finance monitoring. Streaming billing and payment data feed governed KPIs, and Genie answers ad hoc questions against the same figures.
• Inventory and supply chain. Continuously updated stock and shipment tables power dashboards that reflect what is happening now, not yesterday.

FAQs

Are AI/BI Dashboards and Genie generally available?

Yes. Both are generally available and integrate with Unity Catalog governed semantics, so dashboards and conversational answers resolve the same metric definitions.

Do I need to copy operational data into a separate reporting system?

No. AI/BI Dashboards query governed Unity Catalog tables directly through SQL warehouses, so you report on the live tables without maintaining a separate reporting extract.

How do dashboards stay current?

Streaming ingestion keeps the underlying tables updating as new data arrives, dashboards update their charts and filters accordingly, and you can schedule refreshes so viewers land on fresh results quickly.

How do I keep real-time numbers consistent across tools?

Define each KPI once as a Unity Catalog metric view. Dashboards, Genie, SQL, and external BI tools then resolve the same governed definition, so the same metric reports the same value everywhere.