Open a semantic model and you land in its workspace: one header that carries the model's name and identity, one row of tabs, and the side panels you already use. This page is the tour of that shell; what each tab does is covered with the feature it belongs to.
Open a model from the workspaces panel or the home page. It opens on Diagram, and the address bar names the model and the open tab in the shape /, so a link that does not name a tab lands on Diagram too. While the model loads, the page reads Loading semantic model; an address with a model that does not exist opens the DataPrism home page instead.
The tab row reads:
| Tab | What it shows | Opens when |
|---|---|---|
| Diagram | The model's BI tables and the relationships between them | Always |
| Preview | The model's BI tables and a look at their data | The model has at least one BI table |
| Dependencies | The reports that use the model | At least one report uses the model |
| Docs | The model's documentation | The model is saved |
| Health | Findings across measures, tables and relationships | The model is saved and you can edit it |
| Row-level security | The groups that limit which rows each viewer sees | The model is saved and you can edit it |
| History | The change history of the model | The model is saved and you can edit it |
Point to a tab to see what it does, like Relationship diagram for this semantic model. A tab that is waiting for something explains itself instead: Add a BI table to preview this semantic model, No reports use this semantic model, Add a BI table to configure row-level security. A tab does not show as unavailable while its own data is still loading.
In the Ungrouped placeholder, which gathers BI tables that belong to no model, Docs, Health, Row-level security and History are not offered; see Concepts and glossary. Each tab hands over to its feature page: Diagram and Semantic models and BI tables for the model's tables, Check, document and trace a model for Dependencies, Docs, Health and History, and Row-level security and View as for the security groups. Edit access decides most of the row; Who can do what in DataPrism lists the permissions.
If someone else changed the model while you were editing its name, DataPrism asks you to Reload their version or Overwrite it with yours.
While a Model Builder works on the model, its banner sits above the workspace; Model AI covers what the banner offers.
Two controls sit in the header:
Open a report, for example from Dependencies or from the reports list in the model's details, and the report builder takes over the shell.
The report header folds its action row as the window narrows: the secondary buttons first lose their labels, then fold under More actions, and only at the narrowest width do pinned controls follow; the lowest-priority action goes first. Controls like undo and redo never fold. An active action shows a dot when it is an icon, and a busy action swaps its icon for a spinner.
The More actions menu always carries Rename, Make a copy, Move and Delete, whatever the width, and its Rename opens the same in-place editor a double-click opens. For moving and copying, see Move and copy items. A model's own header keeps its small set of actions in a fixed row.
If the role behind the model is deleted, a banner sits above the workspace: The owner role of this semantic model has been deleted. Queries, previews and reports built on it will fail until it is recreated under an active role. The model's rows and cards show an Owner role deleted badge, and its details add Queries will fail until this model is recreated under an active role. To use the data again, create a new semantic model under a role that is still active; the owner role cannot be changed after creation. See Concepts and glossary.
Diagram. A model opens on its relationship diagram, and a link that does not name a tab lands there too.
Yes. The workspace reopens the tab you last used. A measure is never restored as the view; it opens as a floating editor over the diagram. A tab that no longer applies falls back to Diagram.
Double-click its name in the header, or press F2 while the name is in focus. Type the new name and press Enter, or press Escape to leave the old one. You need edit access to the model, and a model still showing the placeholder name Semantic model cannot be renamed.
Reload shows the version someone else saved and drops your name change; Overwrite saves your change on top of theirs. Reload when you are unsure what else changed.
Open the chips beside the model's name. Access marks Owner role and Data mode as Locked at creation and shows both values, together with the reports built on the model. To run through another role or mode, create a new model; see Concepts and glossary.
Two lists help. The model's details, opened from the chips, list the reports built on it, and the report name in the header opens a switcher with This semantic model and Other reports, plus a row that opens the model itself. Both just open things; to rename or delete a report, use its More actions menu.
The report header folds as the window narrows: secondary buttons first drop their labels, then fold under More actions, and at the narrowest width even pinned controls follow. Undo and redo never fold. Widen the window and the labels come back.
Use the back arrow. It reads Back to report, Back to semantic model or Back to workspaces; with several places to return to it becomes a menu, and Back to workspaces is always last. The model you are already on is not offered.
Yes. The address names the open tab, so a saved link opens the model on that tab. A tab you cannot open, or one that is still waiting for content it needs, falls back to Diagram; an address with a model that does not exist opens the DataPrism home page.
New report appears for a saved model when you can create reports and the model has at least one BI table. Add a BI table first; see Semantic models and BI tables.