Elevate quality with research-backed signals
Know exactly where your AI outputs stand to support learners and educators who depend on them.
What’s pushed to production will be used in classrooms. That responsibility is not taken lightly. We partner with domain experts to iteratively refine our evaluators so developers can empower educators and learners with tools built around how students actually learn.
Literacy Evaluators
Designed to measure whether a text is grade-level appropriate and what features of that text make it rich for instruction. Scrutinize a text’s qualitative attributes, informational context, and literary complexity based on rubrics from Student Achievement Partners, with data annotation and validation from ANet.
Determines whether AI-generated text is suitable for a grade band and suggests scaffolding that can support instruction of the text.
Feedback Evaluators
Assesses whether AI feedback on student writing is formative, actionable, and supportive of learning. They examine whether the feedback strengthens motivation and sustains productive struggle so students persist through challenge and build lasting understanding. These writing feedback evaluators were developed in partnership with Quill.
Checks whether feedback identifies a real, specific strength in the student’s writing and is accurately and directly grounded in what the student actually produced.
Standards Evaluators
Determines whether AI content is aligned with the intended state academic standard. This helps educators ensure they are using appropriate content for their intended lessons.
Determines whether the content aligns with a chosen state academic standard and provides a fine-grained analysis at the learning components level.