Content Understanding routing: sub-analyzers receive the whole document instead of their segment pages (worked on Nov 2, broken since ~Nov 20)

Motyka Ondřej 0 Reputation points
2026-11-20T22:01:15.9333333+00:00

I’m using Content Understanding (API 2025-11-01) with a classifier/router analyzer (enableSegment: true) that splits a multi-document PDF into segments and routes each category to a custom analyzer via contentCategories[].analyzerId.

Segmentation works fine, e.g. invoice on page 1, delivery note on page 2, another invoice on pages 3–4. But the routed analyzer for each segment receives the whole document (startPageNumber: 1, endPageNumber: 5, full markdown), so the second invoice gets the data of the first one. A few weeks ago the same setup returned only the segment pages.

My questions:

Should a routed analyzer process only the pages of its segment? Is the current behavior a known issue?

If not, is the recommended way to call the custom analyzer separately with the range parameter (e.g. range=3-4) using the page ranges from the router?

Thanks!

Azure Content Understanding in Foundry Tools
0 comments No comments

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.