Make PBIDS Files

October 23, 2019 By Mike Carlo

In October of 2019 Power BI released a new file type, PBIDS. The Power BI Desktop Source (PBIDS) file is a JSON object file that aids users connecting to data sources. In true PowerBI.Tips fashion we have of course, made a tool for that.

Introducing Connections

Today we release the new tool called Connections. It can be found at https://connections.powerbi.tips/ . With this tool you can use our predefined templates or customize one of your own file. To learn more about this sweet sweet JSON editing tool check out the following YouTube Video:

Technical Details

For more information on the Power BI Desktop Source file check out these links:

Announcement of release of PBIDS: https://powerbi.microsoft.com/en-us/blog/power-bi-desktop-october-2019-feature-summary/#pbids
Documentation on PBIDS files: https://docs.microsoft.com/en-us/power-bi/desktop-data-sources#using-pbids-files-to-get-data
Full documentation on Data Source Reference specification: https://docs.microsoft.com/en-us/azure/data-catalog/data-catalog-dsr#data-source-reference-specification

Oct 6, 2026

Where Writeback Belongs – Ep.569

An anonymous mailbag asks whether Power BI should let people act on a number without leaving the report. Tommy Puglia and Mike Carlo put writeback in Fabric apps, keep the report for consumption, and treat translytical task flows as the exception.

Oct 1, 2026

FabCon Barcelona - Feature Draft – Ep.568

Tommy Puglia and Mike Carlo draft a team from the FabCon Barcelona feature list, starting with Fabric app backends and the ontology authoring agent. The picks that follow are pay-as-you-go capacity, the Fabric notebook toolkit, deployment hooks, and text search in semantic models.

Sep 29, 2026

The Right Model for the Task – Ep.567

Tommy Puglia and Mike Carlo use Kurt Buhler and Eugene Meidinger's Tabular Editor piece to decide which model and effort level belongs on a Fabric task. The same hour covers FabCon Barcelona's Fabric app backend, semantic views, and why monthly active agents may matter more than monthly active users.