A manufacturing company wants their Copilot Studio agent to automatically extract the purchase order number, vendor name, and total amount from PDF invoices that customers upload during a chat session. A developer plans to use AI Builder for the extraction. What is the correct sequence to implement this capability?
Select an answer to reveal the explanation.
Short Explanation and Infographic
AI Builder is the point-and-click AI training studio, and Power Automate is the bridge that carries its results back to the agent. You teach AI Builder what an invoice looks like, publish it, then Power Automate hands the model a file and shuttles the extracted fields home to the agent.
Full explanation below image
Full Explanation
The correct architecture for AI Builder document processing within a Copilot Studio agent has three distinct stages. First, in AI Builder, the developer creates a document processing model, uploads sample invoices (typically 5 or more per layout), labels the fields to extract (purchase order number, vendor name, total amount), trains the model, and publishes it. A published model has a model ID that can be called programmatically.
Second, a Power Automate cloud flow is created with the 'Run a flow from Copilot Studio' trigger. The flow receives the uploaded file (or a file URL) as input, uses the AI Builder connector's 'Extract information from documents' action referencing the published model ID, and maps the extracted field values to flow output parameters.
Third, in the Copilot Studio topic, the developer adds a file upload prompt, then a 'Call an action' node that invokes the Power Automate flow, passing the uploaded file content and receiving the extracted values back as topic variables.
Option A is incorrect because Copilot Studio's entity settings are for recognizing entities in conversation text, not for processing uploaded document files. Custom entities work on utterances, not binary file content.
Option B is almost correct in structure but omits the critical step of calling the AI Builder model from within the Power Automate flow using the AI Builder connector. Simply invoking the flow without the AI Builder action step would not perform extraction.
Option C is incorrect because Copilot Studio topics cannot directly make HTTP calls to Azure Form Recognizer (or any external service). Even if they could, bypassing AI Builder for Form Recognizer would require managing API credentials separately and would not use the low-code AI Builder training environment the question specifies.
Exam tip: AI Builder models are always called through Power Automate — they cannot be called directly from a Copilot Studio topic. The flow is the mandatory intermediary layer.