Looking at the documentation, the extracted text should come from the content field.

Documentation: https://docs.optimizely.com/graph/docs/text-extraction-from-media

I checked in a SaaS instance and the extracted text goes into _fulltext there:

{
  _Media {
    items {
      _fulltext
    }
  }
}

(1 attachment from the Slack thread was not included.)

You were asking about PaaS in your question, though. The documentation seems to suggest it goes into the content field.

In PaaS CMS 13 I don't see a Content property.

The docs are not that clear about CMS 12 vs. 13 generations. Only a few pages I've found handle this well.

Maybe, as PaaS follows SaaS, this has changed in the CMS 13 version.

CMS 13 should match SaaS.

Since text extraction is probably considered heavy lifting, I'm wondering if there are limitations depending on the type of index or environment that is running.

I've now seen text extraction work in my local CMS 13 PaaS solution as well. It goes into _fulltext the same way as with SaaS above.

There are a lot of files in this setup that I hoped text could be extracted from. Where it wasn't extracted, that fooled me. The simpler PDF (exported from Word) works though.

Could the Graph documentation be made clearer about CMS 12 vs. 13? If the linked page, which says extracted content is in the content field, only applies to CMS 12, that should be called out because it is the legacy version. There were clear schema changes between 12 and 13 on PaaS, so it may help to make CMS 13 the default in the docs, label it clearly, and link to older docs for CMS 12.

Thanks for the feedback. Documentation improvements are coming up, and this will be taken into account.

Thread answered in another thread: https://world.optimizely.com/forum/developer-forum/cms-12/thread-container/2026/10/does-graph-packages-for-cms-12-extract-text-from-media-files