The Process

Three steps. Same flow whatever mode you pick.

Pick a mode, run the interview, review the report. The same engine handles every assignment type — what changes is whether the student also submits a written piece.

Teacher sets up the assessment

Pick the mode (Content Assessment, Submission Verification, or Mixed). Author the questions and a rubric. Optional video proctoring for live interviews. ~10 minutes from blank page to ready.

Student takes the interview

Adaptive AI conversation built on the rubric. Voice or text. The student explains what they know, walks through their reasoning, and demonstrates understanding live.

Teacher reviews and discusses

A transcript, an AI grade, and a per-question report. Override anything, add a note, then publish. The transcript is something to actually talk about in office hours.

What you get back

A transcript, not a number

Every word the student spoke or typed, every question the AI asked, every grade the AI assigned and the reasoning behind it. Material to actually use in conference.

Per-question rubric grading

For each question: a band against the rubric the teacher built, evidence quotes from the student's answer, and a holistic summary across the whole session.

Authenticity signal (when video proctoring is on)

Two-stage authenticity layer flags reading off-camera, coaching, lip-sync mismatch, and audio shifts. Compound signals only — single-channel observations like “they looked away while thinking” never trigger a concern. Teacher always sees the recording.

Teacher controls the final grade

AI proposes; teacher disposes. Override any band, add a teacher note, choose whether to publish to the student. The conversation is part of the assignment — graded as such.