AI-103 Certification Practice Question #64

Question

You have a Microsoft Foundry project that ingests scanned PDF invoices stored in Azure Blob Storage. Each invoice contains printed fine items and has a table-based layout.
Extracted results are stored as structured JSON and used as grounding data for an agent in a Retrieval Augmented Generation (RAG) solution.
You need to create a single analyzer that meets the following requirements:
Extracts the invoice number, invoice date, vendor name, and total amount across varying templates
Returns confidence scores so that results with confidence below 0.80 can be routed for supervisor review
What should you use?

Accepted Answer

The requirement is a **single analyzer** that (1) extracts four specific business fields — invoice number, date, vendor name, total — across **varying templates**, and (2) returns **per-field confidence scores** so values below 0.80 can be routed for supervisor review. Azure Content Understanding's custom document analyzer is designed exactly for this: you define the required fields with the `extract` method and enable `estimateFieldSourceAndConfidence` to get a 0–1 confidence per field, which MS Learn explicitly recommends using to auto-approve high-confidence results and route low-confidence ones to human review. The prebuilt-layout analyzer (C) returns document structure but not the four named fields; groundedness guardrails (A) measure RAG response consistency, not extraction; and documentSearch + `search.score` (D) yields a search-relevance score rather than a field-level extraction confidence. Suggested Answer, the embedded `✓`, and MS Learn all align on B; confidence is strong, held just under the top band because there is no community vote for independent corroboration.

More AI-103 practice questions