The phrase gets used two ways. Loosely, it covers any model that hands software a decision rather than a paragraph, which includes encoder classifiers like BERT and DeBERTa, span taggers like GLiNER, and anything wrapped in constrained decoding to force a fixed output shape. More narrowly, people use it for the new category TypeSafe is trying to define, where the typed answer and its probability are what the model was trained to produce in the first place.
There is no standards body here. Treating “decision model” as a description rather than a spec avoids arguing about whether a given system qualifies.
What a concrete example looks like: a support ticket goes in as state, and a Choice question with the options shipping, returns and billing comes back as {"choice":"returns","confidence":1.0,"probabilities":{"shipping":0.0,"returns":1.0,"billing":0.0}}. Your router switches on the string. The probabilities are there so you can refuse to act when they are flat.
A Choice accepts up to 255 options, so deep taxonomies need more than one question. Whether this is a new category or a well-packaged classifier is the argument covered in is Jev just a classifier.