What it is
Voice-based competency assessment for technical and vocational education
AkoVoice turns a course rubric into a talking AI assessor. Students scan a QR code at their workbench and speak with the AI for 5–10 minutes. The AI uses Socratic questioning against the qualification rubric, probes incomplete answers, and produces a criterion-level judgement with evidence quoted directly from the conversation. The tutor reviews the session, or moderates the AI's judgement, on a mobile dashboard.
Currently piloting in English, Korean, and Mandarin across welding safety, automotive pre-delivery inspection, and aged care communication. Higher-level engineering and services programmes are joining in semester two 2026.
Why it matters
Designed for the workshop floor
Most TVET assessment is written. But the work isn't. Students demonstrate competency through their hands and their voice — explaining what they did, why they did it, and what they'd do differently. AkoVoice meets students where the work happens, lets them speak in their own words, and gives the tutor a structured, evidence-backed view of what each learner actually knows.
The methodology is open under CC BY 4.0. The rubric specification is jurisdiction-neutral, supporting binary, three-level, and four-level grading frameworks used across different national qualification systems.
Where it comes from
A practitioner project
AkoVoice extends three years of Gen-AI text chatbot practice at a New Zealand Institute of Technology and Polytechnic (ITP), where Cogniti.ai (University of Sydney) has run continuously since 2023. A 2024 study with the Bachelor of Nursing cohort showed 70% of students improved their summative test scores between assessments — compared to almost no improvement in the previous two cohorts. The study was peer-reviewed and published in Nursing Praxis in Aotearoa New Zealand (Mathews, Adams & Cheyne, 2025).
That paper identified voice as the next interface "powering interactive oral assessments." AkoVoice is the response.