Grounding Data Source for Generative AI in Power Platform
Short answer
C Endorse the semantic model to mark it as a trusted, high-quality, reusable data source for Copilot consumption.
Question
• Microsoft Dataverse tables
• A Microsoft Power BI workspace named WS1
• A canvas app named App1 that uses Dataverse
• A Power BI semantic model that connects to Dataverse by using DirectQuery
You plan to use generative AI to provide answers to queries based on a subset of corporate data.
You need to ensure that the data is available as a grounding data source for AI systems.
What should you do?
Key Takeaways
The question tests the governance mechanism (Endorsement) required to expose existing Dataverse-backed data to AI systems, often confused with simply populating the underlying store.
To provide answers to queries based on corporate data using generative AI, you must ensure the data is available as a grounding data source. Endorsing the semantic model marks it as trusted and ready for Copilot consumption.
Common mistake: Candidates often select 'Populate a Dataverse table' because they focus only on the storage layer, missing that the AI system requires a specific governance signal (endorsement) to trust and utilize the semantic model.
Expert Analysis
Why the Answer Is Correct
Endorsing the semantic model (Option C) is the correct action. In Microsoft Fabric and Power BI, endorsement (Promoted or Certified status) signals to AI systems like Microsoft Copilot that the model is a trusted, high-quality, reusable data source. This governance step is explicitly required for the model to be used as a grounding source.
Why the Other Options Are Wrong
Option A (Populate a Dataverse table) is incorrect because while Dataverse is the storage, the AI agent cannot automatically ground on unendorsed data. The prompt implies the solution already exists; the missing piece is the AI readiness configuration. Option B (Share WS1) grants access but does not establish the data's quality or trustworthiness for AI grounding. Option D (Export the semantic model) is irrelevant to exposing data to the AI service.
Community Comment Notes
Community consensus strongly supports Option C. As user 'hi1011' noted, Microsoft Learn documentation suggests using tags to label models as ready for Copilot consumption, which aligns with endorsement. User 'VincentVega33' clarified that while A provides the data, C provides the necessary trust signal for AI to use it.
Exam Strategy
When dealing with 'AI grounding' or 'Copilot' questions in the AB-100 exam, look for governance actions like 'Endorse', 'Certify', or 'Tag'. Do not assume that raw data availability (populating tables) is sufficient without the proper trust indicators.
Frequently Asked Questions
Why isn't populating the Dataverse table enough?
What is the difference between Promoted and Certified?
Discussion 10 contributions
Compiled from public community threads and reader submissions, lightly edited for clarity. Handles and vote tallies are shown as recorded, and comments reflect the views of individual contributors.
From public community threads 10
-
hi1011 👍 3 Answered C"Consider using tags to label semantic models as ready for Copilot consumption. This method helps data consumers identify models that they can use with Copilot and expect better results. Alternatively, consider the readiness of a semantic model for use with Copilot as a criteria for its endorsement to promoted or even certified status." https://learn.microsoft.com/en-us/power-bi/create-reports/copilot-semantic-models
-
deca0d4 👍 3 Answered CTo endorse a semantic model means marking it as a trusted, high-quality, reusable data source in Power BI/Fabric.
-
omath 👍 2 Answered AThe correct answer is: A. Populate a Dataverse table. To make data available as a grounding data source for generative AI systems (including Copilot Studio, Microsoft 365 Copilot, and other AI experiences), the data must live in a supported grounding store. From the options provided, the only store that AI systems can directly ground on is: Microsoft Dataverse The scenario already includes Dataverse, so the correct action is to populate a Dataverse table with the subset of corporate data you want to expose to AI.
-
moongang 👍 1 Answered ADataverse is the native grounding data source for Copilot and other generative AI features in Power Platform (Copilot Studio knowledge sources, Copilot in model-driven and canvas apps, AI Builder prompts). Loading the relevant subset of corporate data into a Dataverse table makes it directly available for retrieval and grounding, with Dataverse security roles controlling what the AI can see. Why the others don’t fit: Share WS1 only grants users access to the workspace; it doesn’t expose the data to AI systems as a grounding source. Endorse the semantic model (promote/certify) signals quality and trust for reuse in Power BI, but endorsement by itself doesn’t make the data a grounding source for the Power Platform AI scenario described. Export the semantic model moves the data out of the platform entirely, which is the opposite of what you want.
-
Shafz007 👍 1 Answered AEndorsing the semantic model marks it as promoted or certified, helping users identify trusted content and enabling discoverability. It does not itself configure the model as an AI grounding source. Microsoft Learn So A. Populate a Dataverse table remains the best answer. The required data subset can then be configured as a Dataverse knowledge source. My earlier explanation needed that distinction: populating the table prepares the data; adding it as a knowledge source enables grounding.
Show 5 more comments
-
VincentVega33 👍 1 Answered Csince its either A or C - A. A -Populate a Dataverse table Simply storing data in Dataverse does not make it a grounding source for AI. The AI needs a supported, trusted data source such as an endorsed semantic model. Wheras C - Endorsing a semantic model (for example, marking it as Certified or Promoted) signals that it is a trusted, curated source of business data that AI can use to answer questions accurately.
-
kinser123 👍 1 Answered CSemantic Model Endorsement (Promoted or Certified): In Power BI and Fabric, endorsing a semantic model explicitly designates it as an authoritative, verified, and trusted dataset. Generative AI grounding mechanisms rely on endorsed/certified semantic models to ensure the AI retrieves validated corporate data rather than unverified sources.
-
AxaptaKing1 👍 1 Answered CAnswer in this case is C but Its important to understand, Semantic Model Endorsement is not grounding itself, but it is a very useful governance mechanism that helps AI systems choose and rely on trusted data sources for grounding.
-
Luweho 👍 1 Answered CA semantic model can (and actually should) be used as a grounding source, preferred over raw dataverse tables that the AI agent might misinterpret. The "subset" requirement is not directly addresses here, but also not in A.
-
Neoturka 👍 1 Answered AYou need to ensure that the data is available as a grounding data source for AI systems.
Reader comments & corrections
No reader corrections yet — spotted an error or have a note? Share it below.