Dashboard migration moves from one file to a folder

Databricks has expanded the Genie Code importBI skill so teams can migrate a folder of Tableau or Power BI reports in one run. Users place source files in a Unity Catalog volume, open a draft AI/BI Dashboard and invoke the skill. Genie Code discovers reports, pairs workbooks with data sources and creates dashboards and metric views in a chosen catalogue and schema. The new batch path targets organisations with large business-intelligence estates where converting one report at a time would make modernisation slow and inconsistent.

Batch visibility gives operators a control point

Reports are queued and processed in batches, with a status card showing whether each item is queued, running or completed. A user can stop one report or the entire batch. These controls matter because a migration agent may encounter duplicated sources, unsupported calculations or a report that expands into an unexpectedly large workload. A stop button should leave the system in a known state, identify partially created assets and allow a clean retry. Operators need downloadable logs rather than relying on a temporary interface for diagnosis.

Reuse can reduce duplicated semantic work

Genie Code attempts to reuse existing Unity Catalog tables, views and metric views. Reuse can prevent every migrated dashboard from defining revenue, customer or margin differently. It can also connect a new report to the wrong asset if names appear similar while their filters or grain differ. Matching should surface evidence and confidence, with a review path before shared semantic objects are changed. Organisations should establish certified metrics first where possible, then let the agent map reports to those definitions rather than treating generated metric views as automatically authoritative.

Visual conversion is only the visible layer

A dashboard may look similar while calculating a different answer. Tableau and Power BI contain proprietary expressions, filters, relationships, row-level security and interaction patterns. Migration validation must compare data and behaviour as well as layout. Teams should test representative parameter values, drill paths, totals, null handling and date boundaries. Pixel-perfect recreation may not be necessary, but any intentional difference needs documentation. The business owner should confirm that the converted dashboard supports the decisions for which the original was built.

Permissions must survive the move

Source dashboards often rely on workspace roles, embedded credentials and row-level rules. Unity Catalog provides a governed target, but the migration needs an explicit mapping from old identities and groups to new privileges. A report that becomes broadly readable after conversion creates a data incident even if every chart is accurate. Test accounts with different roles should open both versions and compare visible rows and available actions. Generated assets also need clear owners so permissions do not remain attached to the migration operator indefinitely.

The semantic layer is the durable outcome

Databricks encourages teams to promote generated metric views into Unity Catalog or request governed metric views during migration. That can make the project more valuable than a format conversion by turning repeated calculations into reusable definitions for dashboards and agents. Promotion should require review, naming standards, tests and lifecycle ownership. A generated metric that serves one report may not be suitable as an enterprise definition. Lineage back to the source report and data set helps reviewers understand why a calculation exists and which consumers could be affected by a change.

Prioritisation prevents a migration factory from moving clutter

Large BI estates contain unused, duplicated and obsolete reports. The ability to migrate a folder should not encourage teams to convert everything. Usage logs, owner interviews and regulatory requirements can identify which dashboards deserve investment. Some reports should be retired, consolidated or replaced with a simpler metric. A migration programme should measure active adoption and decision value, not the number of files processed. Genie Code is most useful when it accelerates a curated plan rather than preserving years of dashboard debt in a new platform.

Agentic conversion still needs an acceptance discipline

Bulk migration can shorten mechanical work and make progress visible, especially when source files follow consistent patterns. It does not remove the need for business and data owners. A credible acceptance process compares outputs, permissions, performance and refresh behaviour before replacing the old dashboard. Databricks has added useful operational controls and a route to governed semantics. The strongest outcome will be a smaller, trusted AI/BI estate with reusable metrics, not merely a high count of automatically converted reports that users must later rediscover and repair.

Performance deserves a separate acceptance gate. A converted dashboard may be logically correct yet too slow for interactive use because its query pattern, caching or source model differs from the original platform. Teams should compare load time and warehouse consumption under realistic concurrency, then tune metric views rather than hiding latency behind larger compute. Usage telemetry after launch can reveal filters or reports that need redesign. This evidence keeps migration success connected to user experience and operating cost.