For the past two years I’ve been working with my colleague Inder Rana to build out a GitHub repo that can be leveraged by Fabric users for training and testing purposes. In the most recent update we have added a new Fabric Data Agent that works with the semantic model using Direct Lake mode. The new AI Instructions for the Direct Lake Semantic Model will also work with Power BI Copilot (see examples for Power BI Copilot standalone version in the video below!). Alongside the existing Lakehouse/Warehouse Fabric Data Agent, you can now compare and contrast both versions.
A video reviewing the new Semantic Model Fabric Data Agent capabilities in the GitHub repo can be watched below or by clicking this link:
The updated GitHub repo that deploys 275 million rows of real healthcare data to your Fabric environment for an end-to-end demo and testing solution can be found at this link: fabric-samples-healthcare/analytics-bi-directlake-starschema at main · isinghrana/fabric-samples-healthcare
If you have already deployed the GitHub repo, you can add the new Semantic Model Data Agent by following these instructions: https://github.com/isinghrana/fabric-samples-healthcare/blob/main/analytics-bi-directlake-starschema/docs/5-CreateAISkill.md#optimize-semantic-model-for-power-bi-copilot-and-fabric-data-agent
The GitHub repo can deploy all of the following Fabric components in a single quick install script:
- Lakehouse
- Notebooks
- Data Factory Pipeline
- Semantic Model (optimized for Power BI Copilot and Fabric Data Agents)
- Power BI Report
- Data Agent for Lakehouse
The following components can be deployed manually after running the quick install script:
- Warehouse
- Data Agent for Semantic Model
I previously recorded a video to review best practices for Semantic Models with AI for my YouTube channel which can be viewed below or by clicking here:
Leave a Reply