Grounding an AI Hub prompt with Dataverse in Power Automate
Short answer
C Select Dataverse table columns as the grounding knowledge source so the prompt retrieves live business data during the Power Automate run.
Question
The prompt must use Retrieval-Augmented Generation to ground the model with additional business information.
You need to select the available knowledge source for the prompt.
What should you use?
Key Takeaways
It tests the RAG grounding step in AI Hub Prompt builder; the trap is picking a static artifact such as a PDF or plain text, which supplies no run-time retrieval.
This AB-410 prompt-building question tests which knowledge source supports Retrieval-Augmented Generation for a prompt used in Power Automate. The correct grounding source is Dataverse table columns (C), retrieved live at flow run time.
Common mistake: Option D (Text) is most tempting because text contains the specifications, but pasting data into the prompt is not RAG — nothing is retrieved at run time. Option A (PDF) is likewise a static file, not the supported Dataverse grounding source.
Expert Analysis
How the prompt gets grounded The question traces a single path: a custom prompt authored in AI Hub (Prompt builder) is called from a Power Automate cloud flow, and before the model runs, the prompt must be enriched with live business context through Retrieval-Augmented Generation. In AI Builder, RAG is implemented as a grounding step: you add a knowledge source, the runtime retrieves the relevant rows/columns from that source at execution time, and those retrieved values are injected into the model's context alongside the prompt text. So the deciding factor is which option is a supported, structured knowledge source that the prompt runtime can query.
Why the answer is C Dataverse table columns are exactly that source. When you configure grounding in Prompt builder you select one or more Dataverse tables and the specific columns to retrieve; the flow's run then pulls matching rows (including related-table columns) and passes them as context. Because Dataverse is the system of record for Power Platform, this retrieval happens live at each flow run rather than freezing information into the prompt definition — which is the whole point of grounding versus static prompt text. Correct answer: C.
Why the other options fail A PDF document (option A) is a static file: it would require upload/OCR ingestion and is not the supported Dataverse grounding source for a prompt used in Power Automate. A PNG image (option B) is not a knowledge source at all — the retrieval step reads tabular business data, not image content. Plain text (option D) is simply part of the prompt itself; putting information in the prompt is not RAG, since nothing is retrieved at run time. The trap is choosing text or a document because it 'contains information' — RAG requires an external source queried during execution, and in this stack that source is Dataverse.
Exam Strategy
When a question says the prompt must use RAG or grounding, look for the option that is a queryable, structured system of record — in Power Platform that is Dataverse. Eliminate any option that is static content pasted into or uploaded with the prompt (text, files, images), since retrieval must happen at run time. Also check the calling context: 'used in Power Automate' points to AI Hub prompts grounded on Dataverse rather than SharePoint or website sources used elsewhere.
Frequently Asked Questions
Why can't I use a PDF file as the knowledge source for a Power Automate prompt?
How does grounding differ from simply writing product data in the prompt?
Community Votes
100% of anonymous learners picked answer C. Votes are pick records left by other test-takers — they are not the verified answer.
Discussion 2 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 2
-
aed747e 👍 3 Answered CIn AI Builder, custom prompts use Retrieval-Augmented Generation (RAG) by configuring Dataverse grounding. Grounding links the prompt directly to live enterprise data stored in Dataverse table columns (and related tables), supplying necessary business context to the AI model before the prompt executes inside a Power Automate workflow.
-
MeshMallow 👍 1 Answered CMicrosoft Prompt Builder supports Retrieval-Augmented Generation (RAG) by using Dataverse tables as knowledge sources.
Reader comments & corrections
No reader corrections yet — spotted an error or have a note? Share it below.