AI/BI Dashboards

[km1837]

Hi All

My client is looking to create a few dashboards in DataBricks instead of Power BI. We already have a template in Power BI (header for each page, information about out BI Page) in each of our power bi report which has colors, fonts, images etc.

How do I replicate the same in our DataBricks dashboards. We want to create a template first and then start creating actual data dashboards after that.

Also is there a way to put different workspaces for Dashboards, like one for Operations, Finance and Sales. We are BI Team and we serve all business. 

#dashboards

[srini_ve]

@km1837 
A practical approach would be to use the existing Power BI template (.pbit) as the starting point rather than trying to recreate the design from scratch.
Since you already have a standard Power BI template containing the page header, colours, fonts, images and other common branding,
I would suggest the following approach:

Step 1: Create/export the standard Power BI template as a .pbit file.
Step 2: Import the .pbit template when creating the Databricks dashboard template.
Step 3: Validate the imported components from power bi like below.
Page headers
Colours and fonts
Images and logos
Page layout
Visual formatting
Other standard branding elements

Some Power BI-specific components may need adjustment or recreation depending on their compatibility with Databricks dashboards, so this validation step is important.

Step 4: Adjust the Databricks dashboard template where required.
Step 5: Get business approval.
Step 6: Start building the actual dashboards.

Once the template is approved, the BI team can use the same standard design as the starting point for the actual dashboards. Only the dashboard-specific data, KPIs, charts and filters would then need to be added.

This approach should help us maintain a consistent look and feel across all Databricks dashboards while reducing the effort of recreating the same branding and layout repeatedly.

For Workspace organisation

Databricks Workspace

* Operations Dashboards
* Finance Dashboards
* Sales Dashboards

Within each area, the relevant dashboards can be grouped together and access can be controlled based on the business users.

For example:

* Operations users → access to Operations dashboards
* Finance users → access to Finance dashboards
* Sales users → access to Sales dashboards
* BI Team → access across all areas

The important part is to manage access using the appropriate Databricks groups and permissions, rather than giving all users access to everything.

[balajij8]

@km1837 

You can start with Workspace Themes - admin level configuration that defines fonts (family, color, size, weight per text category), canvas background, widget styling (border, corner radius, padding), visualization axis/grid colors, and categorical/gradient color palettes including color mappings that assign specific colors to named values across all charts. Every new dashboard inherits this theme automatically, covering the majority of the branding requirements. For the header banner with logos and titles that you use in Power BI, combine Text widgets (rich text with headings, links, inline images, tables) and Image widgets (standalone logos or branded visuals from URLs or Unity Catalog volume paths) placed at the top of each dashboard page. You can also upload custom thumbnail images per dashboard for visual identification in the dashboard listing.

You can build a template dashboard with your workspace theme applied, header widgets positioned and standard layout structure and export it as a json file via the menu (File actions > Export) to create a reusable template. Each time your BI team starts a new dashboard, import that file to get a pre-branded starting point and then swap in the actual datasets and visualizations. Store the template in a Git folder for version control and use DABs to deploy across dev/test/prod environments, giving you a CI/CD workflow comparable to Power BI deployment pipelines. Alternatively, since new dashboards auto-inherit the workspace theme, your team can skip the import step and manually add the header text and logo image widgets letting the theme handle all colors and fonts automatically.

For organizing dashboards by business domain (Operations, Finance, Sales), use workspace folders with folder-level permissions ( /Workspace/Shared/BI-Dashboards/Operations/ ) - so each team sees only their area while sharing the same workspace theme and compute. Layer on Domains, which group dashboards, tables and Genie Agents by business area on the Discover page, giving consumers a browse-by-domain experience. Avoid creating separate Databricks workspaces per business unit unless you have strict regulatory or administrative separation requirements as it adds operational overhead and cost that a central BI team typically does not need.

[Coffee77]

I would suggest you standardize Databricks dashboards, although the approach is a bit different from Power BI.

For the visual template, I would create a “master” Lakeview/AI/BI dashboard containing the common elements: header, logo/images, standard text, colors, fonts where configurable, filters and the general page layout. Then use that dashboard as the starting point for new dashboards by cloning/duplicating it. Keep in mind that Databricks does not provide the same level of pixel-perfect theming and visual customization as Power BI, so some elements of your existing Power BI template may need to be simplified.

For organization, I would not create separate Databricks workspaces just for Operations, Finance and Sales dashboards unless you need strong isolation between those domains. A workspace is a relatively high-level isolation boundary and create isolated ones only for dashboards does not make sense from my view. I would only introduce multiple workspaces if there are additional requirements around isolation, administration, environments, compliance or different teams that really justify that extra complexity.

Instead, I would keep the BI dashboards in the same workspace and organize them using folders, for example:

/BI
 /Operations
 /Finance
 /Sales
 /Templates

Then combine this with appropriate permissions for each folder/dashboard.

For the data itself, Unity Catalog should provide the governance layer. You can organize catalogs/schemas according to your domain architecture and grant the appropriate access to Operations, Finance and Sales.

Besides, if you have multiple environments (DEV, QA, ..., PROD), you can even export/import underlying code to deploy dashboards using Asset Bundles.

I hope it helps.

[Ashwin_DSA]

Hi @km1837,

Databricks AI/BI Dashboards have solid theming and organisational capabilities that should cover most of what you are used to in Power BI.

AI/BI Dashboards don't have a "template file" concept the way Power BI does with .pbit files, but you can achieve the same consistency through two features (Workspace themes and Text widgets with images) working together.

Workspace Themes are the closest equivalent to a Power BI template for branding. A workspace admin can define a default theme that controls fonts (family, colour, size, weight, and case per text category), canvas background colour, widget styling (background, border, corner radius, padding, margin, shadow), visualisation colours (axis, grid lines, categorical and continuous colour palettes), and textbox alignment. Every new dashboard in the workspace automatically inherits this theme, so once it's set up, your team gets consistent branding without any extra effort. You configure it under Settings > Appearance > AI/BI > Theme. See the example below. 

Text widgets with images handle the header/banner part of your Power BI template. You can drag a text widget onto the canvas and use the formatting toolbar or raw markdown to add your company logo (via image URL or path), styled header text, links, and tables. This gives you the equivalent of a branded header on each page.

For a practical "template" workflow, I'd suggest: set up the workspace theme for your brand colors and fonts, then build one dashboard with your standard header layout, page structure, and any boilerplate text widgets. Export it as a .lvdash.json file (kebab menu > File actions > Export), and use that as your starting point for new dashboards by importing it and swapping out the data. You can even edit the JSON directly for bulk changes.

With regard to your question about organising by departments, You don't need separate workspaces for Operations, Finance, and Sales. Within a single workspace, you can organise dashboards into folders in the workspace browser, just like you would with any file system. Create a folder structure like /Dashboards/Operations, /Dashboards/Finance, /Dashboards/Sales and manage access through folder-level permissions. 

If your organisation does want stronger isolation (separate compute, separate access controls at the workspace level), you can use multiple workspaces and transfer dashboards between them using the export/import workflow or the Workspace API. But for most BI teams serving multiple departments, folders with permissions within a single workspace are simpler and easier to manage.

By the way...you can now import .pbit(Power BI template) files directly into Genie Code, and it will build an AI/BI dashboard that replicates your visualisations. It also creates metric views that mirror the underlying business logic and detects relationships from the source file. This is a great option for migrating existing reports, but there are a couple of things worth noting for your scenario. The import focuses on recreating the data model and visualisations, not the branding layer (headers, logos, colour themes). So even after importing, you'd still want to set up a workspace theme and add your branded text widgets. It also requires partner-powered AI features to be enabled on both the account and workspace.

The workflow is straightforward: go to Dashboards > Create > Import from Power BI, attach the .pbit file, and Genie Code builds the dashboard.

You can also use the /importBI command within Genie Code if the file is stored in a Unity Catalog volume.

Hope this helps.

If this answer resolves your question, could you mark it as “Accept as Solution”? That helps other users quickly find the correct fix.