Question

A retail company's Power Apps canvas app lets store associates capture customer feedback. The solution architect wants the app to automatically categorize free-text feedback into product themes and store the result in Dataverse, without requiring the associate to press an extra button. The categorization requires reasoning over an evolving list of product categories that changes weekly. Which design approach best fits this requirement while keeping the categorization logic maintainable by makers?

Accepted Answer

Call an AI Builder prompt (prompt action) from the canvas app that classifies the text against a maintained category list, returning structured output written to Dataverse

Answer

Embed a static Power Fx Switch() formula in the canvas app that maps keywords to categories

Answer

Build a Copilot Studio autonomous agent that monitors the Dataverse table and reclassifies records nightly

Answer

Train a custom vision model in Microsoft Foundry to interpret the feedback text