GCP Generative AI Leader Practice Question 26: Document AI Invoice Extraction
Google Cloud Generative AI Leader · Test 3 · question 6 of 10
A finance team wants to build an agent that processes hundreds of vendor invoices a day: it should pull out the invoice number, vendor name, line items, and total amount from scanned PDFs so a human only reviews exceptions.
Which Google Cloud API should the agent call as a tool to extract this structured data from the invoice documents?
Show answer and explanation
Answer:
Document AI is Google Cloud's API for parsing documents. Its invoice processor is purpose built to extract fields such as invoice number, supplier name, line items, and total amount from scanned invoices and PDFs, along with confidence scores, which is exactly what the finance agent needs to hand off structured data for review.
Option B, Cloud Vision, detects objects, labels, and text in images generally. It can perform raw optical character recognition, but it does not understand invoice layout well enough to map text to named fields the way a Document AI invoice processor does.
Option C, the Natural Language API, analyzes syntax, entities, and sentiment in already-extracted text. It does not read a scanned document's layout or pull structured fields from it.
Option D, the Translation API, converts text from one language to another. It has no role in extracting fields from a document.
Reference: Document AI documentation