AI Teacher
Training
VR

Classroom management is learned in conversation. Teachers have to respond to different personalities, behaviours and situations as they unfold, and that skill cannot be built from theory alone. Traditional role-play depends on other people being available and never quite reproduces the unpredictability of a real classroom.

Melbourne Polytechnic wanted teachers to practise evidence-based classroom strategies inside realistic conversations, and to know whether those strategies were genuinely being applied, not just talked about.

The answer was AI. We built an experience around a large language model tailored with Melbourne Polytechnic, where teachers speak naturally with virtual students who respond realistically, while the system monitors how the teacher handles them and assesses whether the principles are being put into practice.

We created a virtual classroom populated by AI-powered students that educators can speak to naturally.

Rather than following predetermined dialogue choices, the experience allows conversations to unfold in real time. Virtual students listen to what the educator says, interpret their response and answer according to their individual personalities and the situation taking place.

This turns the training from a scripted simulation into a responsive environment where the learner has to think, communicate and adapt in the moment.

Communication skills are shaped by much more than the words being spoken.

Virtual reality places educators inside the classroom, surrounded by the students they are interacting with and responsible for managing the situation as it develops.

Combined with conversational AI, the experience introduces the unpredictability that makes real classroom interactions challenging, letting learners practise responding under realistic conditions without the consequences of getting it wrong with a real student.

The training is built around five evidence-based classroom management strategies, from positive reinforcement and maintaining expectations to modelling, re-direction and de-escalation. Teachers learn each strategy, see it in action, then practise it live in a classroom scenario with two virtual students in front of them.

Each student has their own personality, behaviour and context. The tailored language model lets them understand natural speech and answer in real time, so the same scenario can take a different direction depending on how the teacher approaches it.

The AI listens both ways. As the conversation unfolds it monitors how the teacher responds to the students, and at the end of each scenario determines whether the strategy was successfully applied. Open-ended conversation becomes measurable practice, and teachers can replay any strategy as often as they like to build confidence.

The experience explores how VR and AI can work together to create a new kind of immersive learning.

By combining the presence of a virtual classroom with students capable of genuine back-and-forth conversation, educators can practise situations that have traditionally been difficult to simulate at scale.

The result is training that feels less like following a scenario and more like having the conversation for real.