Skip to main content

Evidence of learning across your courses.

Each student talks through their own work with an AI agent, in about 5 minutes. Instructors read Verdicts with evidence from the Transcript and decide the grade. When AI can write the paper, the conversation shows what was learned.

Request a demoTake the guided tour

Currently running at Tulane University and in high schools.

Conversation 5 min 12 sec

  1. AI agent0:09You say a right answer isn’t enough. Why?
  2. Jordan0:14Socrates wants the reason behind it. A correct example alone doesn’t show you understand.
  3. AI agent0:24Couldn’t someone memorize that reason, too?
  4. Jordan0:31They could. I’d try a new case and see if the same reasoning still works.
A Conversation as the Instructor reads it beside the Result: the Recording, then every line of the Transcript. An illustration drawn from the app’s own components, for a fictional student.

The same evidence, whatever the course.

Assessment reform asks for tasks that still show what a student understands when AI can write the work. A short Conversation about their own work is one, and its Result reads the same way in every department.

  • An Assignment, for work students wrote

    Each student uploads their paper, and the AI agent asks questions written from it. The Result gives two Verdicts, Understanding and Authorship, with the lines of the Transcript behind them.

  • A Quiz, for a topic

    Every student answers the same questions aloud, and each question gets its own Verdict, with what the student said.

  • Never a grade

    A Result is Verdicts with evidence: no grade, score or percentage. Only Instructors see Results, and they decide the grade.

  • About 5 minutes a student

    Students start from a link their Instructor shares, on their own time, with nothing to install.

Review and Concerns mean look closer.

A Result is never a finding of misconduct. When the Authorship Verdict reads Review or Concerns, we recommend a closer look, and a chance for the student to explain, before any decision.

Authorship

  • Concerns

    Flagged for a conversation.

  • Review

    Worth a closer look.

  • Clear

    Nothing here needs your attention.

The Authorship Verdict at each of its three stops, with the words the app shows beside it. Drawn from the app’s own component.

What happens next

  1. The Result is marked Needs review

    It stays marked in the Instructor’s list of students until they resolve it.

  2. The Instructor reads the evidence

    Each finding, the lines of the Transcript behind it, and the Recording.

  3. The student can explain

    Concerns reads “Flagged for a conversation.” If something still needs explaining, the Instructor asks the student.

  4. The Instructor decides, and records it

    Mark resolved clears Needs review, with a note if they want one. The Result itself never changes, and every resolve and reopen is kept. Students never see the Result or the note.

  5. Your own process applies

    Any question of misconduct goes through your institution’s process, which a Result can inform and never replaces.

Integrity signals are context, not proof.

For integrity offices: during a Conversation the app notes when a student leaves the tab, tries to copy text or moves the pointer outside the conversation. Each can happen for an ordinary reason, so the Instructor reads them with the Transcript.

  • Students know first

    Before they start, students accept three rules: keep the pointer inside the conversation, stay on this tab, and copy and paste are off. They are told their Instructor sees a note if they leave the tab, move the pointer outside the conversation or copy text.

  • Counts beside the Result

    The Instructor sees how often each happened, with a short summary. Three or more tab switches, or any copy attempt, mark the Result Needs review.

  • None on an open-book Quiz

    An open-book Quiz runs without these checks.

Law schools and business schools

Business schools: assurance of learning

Assurance of learning, as AACSB accreditation uses it, looks for direct evidence that students meet a program’s learning goals. Set a Quiz on a goal and every student in the course answers the same questions aloud. Each Result gives a Verdict for every question, with what the student said as the evidence.

Law schools: formative assessment and the Socratic method

The Socratic method questions students about their own reasoning, but in class only a few are called on. In an Assignment, every student talks through their own memo or brief with the AI agent, which asks questions written from it. Set early in the term, it gives each student formative assessment, and their Instructor the evidence for feedback.

For IT, privacy and procurement

What reviewers ask first, with the rest on one page.

Trust: privacy, retention, sub-processors and security

  • Your LMS and sign-on

    Works with your LMS: Canvas, Google Classroom and others. Single sign-on: Google and Microsoft sign-in, and others. Tell us which you use when you request a demo.

  • Nothing to install

    Instructors and students use OralExam.ai in the browser. Students start from a link.

  • What is kept, and for how long

    Recordings are deleted 90 days after they are made. The work students upload, their Transcripts and their Results are deleted 12 months after the term ends.

  • Agreements

    For an institution, terms are agreed with you, with a data processing agreement and a security review.

Start with one course.

Request a demo and we will show a Conversation and its Result to you and the colleagues you choose. Then we can set up an account so an Instructor can try it with their class.

Email contact@oralexam.ai to request a demo.

Take the guided tour

Teaching a course yourself? Request access