Sentence-level scoring

Sentence-level scoring is when an AI detector returns a probability score for each individual sentence in a document, in addition to (or instead of) a single doc-level number. It's useful when you want to highlight specific sentences for editorial review, downrank parts of a mixed document, or surface exactly which passages look AI-generated.

AI Detector API returns sentence-level scores by default on every /v1/detect call. The response includes a sentences array where each item has the sentence text and its individual confidence score.

Sentence-level scoring is especially valuable for LMS and editorial workflows, where the question isn't just "is this AI" but "which parts should the teacher or editor look at first." See the LMS use case for an integration example.