This workflow is very simple - you have an text body that you want to read, analyze and create some freetext tags for.

It uses the OpenAI interpolator module that reads the text and then generate X amounts of tags that are create as terms and set on the field. As long as you have added that entity references are allowed to be created on your taxonoy field it should be possible.

The end-result is that you can take 1000's of articles and have them free tagged within 1 hour. Pretty neat.