Hear what your people know, not what they can tick.
Multiple-choice tests measure recall. An oral exam shows whether someone can explain, reason and apply what the role needs. iTutor runs that exam by voice: an AI examiner asks questions drawn from your competency framework, follows up on each answer, scores every requirement and flags signs of AI help, in English, Arabic or up to 12 languages.
How do you run an AI oral exam with voice for employee assessment?
An AI oral exam is a spoken assessment in which an AI examiner asks the candidate questions by voice, listens to the answers, asks follow-ups and scores the responses against a rubric. For employee assessment the rubric should come from the role: the competencies and proficiency levels the organization expects. Because a candidate could relay the questions to another AI, a credible oral exam also needs proctoring that detects outside help.
iTutor (itutor.study) runs voice-based oral exams for employees against your competency framework. The examiner works from the requirements of the candidate's role and level, adapts its follow-ups to each answer, and scores every requirement separately. Anti-AI checks include codes shown only on the candidate's screen (on Growth and Enterprise), a trap for relayed AI answers, voiceprint comparison, and camera and browser checks; a result that looks AI-assisted is held for human review. English voice on L&D Starter ($17 per seat a month), English and Arabic on Growth ($22 per seat a month), 12 languages on Enterprise ($27 per seat a month).
Why written tests miss what the job needs
Most employee assessments are multiple-choice quizzes. They are easy to mark and easy to pass: they test whether someone recognises the right answer, not whether they can explain a procedure to a customer, defend a decision or handle a follow-up question. Managers know this, so the result carries little weight.
Oral exams test the real thing, but they cost an assessor an hour per candidate, vary from one assessor to the next, and are now easy to game with a phone and a chatbot. Running them across a whole company has not been practical.
How it works
Map the role
Build your competency framework in the editor, or import it from an existing PDF on Growth and Enterprise: roles, levels and the requirements each level must meet.
Set the exam
Choose the level and the language. On Enterprise you can also set each role and level to technical or behavioural questions, or to a scenario or conversation exam.
The candidate talks to the examiner
In the browser, with camera and microphone on, the candidate answers spoken questions. The examiner follows up where an answer is thin and moves on once a requirement is covered.
Review the result
You get a score for every requirement, the full transcript and the proctoring events, and you decide on the outcome and the certificate.
What the oral exam gives you
Questions from your framework
The examiner works from the requirements of the candidate's role and level, so every question maps to a competency you defined.
Adaptive follow-ups
Each answer shapes the next question: a vague answer gets a probe, a complete one moves the exam on. Questions are not repeated within a session.
A score per requirement
Results are reported requirement by requirement, with the evidence from the transcript, so you can see where a candidate is strong and where they need training.
Anti-AI proctoring
On-screen code checks (on Growth and Enterprise), a trap for relayed AI answers, voiceprint comparison, and camera and browser checks. A result that looks AI-assisted is held for human review instead of passing.
Arabic and English voice
Candidates can sit the exam in English or Arabic on Growth, and in any of 12 languages on Enterprise.
Certificates per level
A candidate who passes a level earns a certificate for it, optionally after a reviewer approves, with a public verification page.
Pricing: from $17 per seat a month
L&D Starter ($17 per seat a month) runs oral exams in English for a pilot team, with camera, voiceprint and browser proctoring and 2 exams at a time. L&D Growth ($22 per seat a month) adds Arabic voice, on-screen code checks, PDF framework import and 20 exams at a time. L&D Enterprise ($27 per seat a month) adds all 12 voice languages, scenario and conversation exam modes, 100 exams at a time, white-label certificates and SSO.
FAQ
Can the oral exam run in Arabic?
Yes. On L&D Growth the examiner speaks English and Arabic; on Enterprise it covers all 12 languages the platform supports.
How do you stop candidates using ChatGPT during the exam?
Several checks run together: codes shown only on screen that the candidate must enter (on Growth and Enterprise), a trap that catches relayed AI answers, voiceprint comparison with the start of the session, and camera and browser checks for phones, extra screens and leaving the exam window. Flagged sessions go to a human reviewer.
Does the AI decide who passes?
The AI scores each requirement and proposes an outcome. Sessions with integrity flags are held for review, and certificates can be set to need a reviewer's approval before they are issued.
Where do the questions come from?
From your competency framework: the requirements of the level being assessed. You can build the framework in the editor or, on Growth and Enterprise, import it from a PDF.