A new recruit steps into a digital recreation of a rail tunnel and is guided through the evacuation procedure, asking questions as they go. The guide checks their understanding, reacting to every answer and decision. In another experience, a bank employee practises a difficult conversation with a distressed customer who responds in real time. 

Neither guide is scripted, and neither is human – both are AI, brought to life inside a hyper-realistic immersive worlds. This is where immersive technology meets AI, and it’s transforming how people participate with it.

Almost every organisation now has an AI strategy, and almost all of them look the same – a chat window and a text box. Useful, but flat. You type, it types back. The intelligence behind it is remarkable, but the experience asks very little of you and too often changes very little about how people actually work.

In Immersive AI, the user is inside a custom environment with unique characters and objects, powered by AI that gives them context and interaction that feels real. At Catalyst we’ve built an Immersive AI application that helps athletes improve technique through real-time feedback from a Large Language Model (LLM), and an experience where users hold real-time conversations with avatars powered by a custom LLM. They don’t just receive information, they participate in a situation. And participation is what changes behaviour.

1. Powered by custom AI: your IP, your competitive advantage

Off-the-shelf training treats every organisation the same – a generic avatar reading a generic script in a generic room. It demos well but is quickly forgotten.

We build Immersive AI experiences powered by large language models and computer vision, each one modelled and trained on the client’s own world – their domain, expertise, values and frameworks. That specificity is what makes it engaging, immersive and developmental.  And it means the capability that develops strengthens the client’s competitive advantage, rather than teaching generic skills any competitor could buy off the shelf.

Believability drives engagement and learning. When an experience feels real and true to a person’s world, they respond as they would in real life, and a real response is what changes behaviour. Because each experience is built on the organisation’s own values and frameworks, every use reinforces “how we do things here.” Culture isn’t what’s written on the wall, it’s what people do under pressure and when staff practise the values aligned response in a safe, immersive setting, it becomes second nature. Everyone trains to the same standard, so behaviour becomes consistent at scale, not dependent on who ran the induction.

Beyond individual feedback, the AI also detects trends across all participants, what are the common mistakes, behaviour patterns, and where the biggest gaps. All this feeds back into both the program and organisational outcomes.

2. Characters and environments real enough to matter

An experience is only as convincing as the character delivering it and the world it happens in. We’ve had Emergency Services teams tell us they switch off the moment American police cars turn up in a simulator built for an Australian context… the details matter.

This is where Catalyst’s character and environment craft comes in. Our character development pipeline is IP built over hundreds of projects, recognised at the Cannes Film Festival and the Australian Design Awards. Our characters don’t just look exceptional, they’re optimised to perform across devices and headsets, and built to gesture, pause and move in a way that reads as human. Pair that with live AI and they are no longer replaying a fixed animation – they respond to you in the moment, with whatever warmth, tension or authority the scene needs.

The world around them matters just as much. Every environment is built specifically around the world the client operates in, to create true presence – drawing on our experience creating hyper-realistic worlds for projects like Limitless with Chris Hemsworth and Brave VR. The goal is embodied learning – when the body believes it’s in a situation, the learning encodes and lasts longer. A hazard you physically stepped around, a conversation that raised your pulse, a procedure your own hands carried out – these stay with people in a way a slideshow never will.

Because the situation feels real but the stakes aren’t, people can get it wrong safely. A frontline banker can fumble a hard conversation, a new manager can misjudge a coaching moment, and the only cost is a lesson learned. Rehearsing the hardest moments in advance replaces “I’ve never done this before” with “I’ve done this a hundred times”. This helps protect wellbeing as much as it builds capability.

3. An immersive training platform, not a single moment

Immersive AI is built on a platform designed to change behaviour through five stages:

1. Why it matters. Opens with the real stakes.  The incident, the lost customer, the personal cost – so people care before they learn.
2. Show the framework. The AI guides the user through the client’s own process or approach, not a generic model.
3. Practice. The user applies the framework with the AI alongside them.
4. Assessment. The user completes the task while the AI measures performance.
5. Feedback. Rather than a quiz score, the user gets real-time, personalised insight into how they did, where they hesitated, and how they compare to best practice and their peers. All built on the client’s own knowledge and standards.

What this could look like

Infrastructure. A worker steps into an immersive safety induction and meets Fred, an AI safety guide trained on underground rail safety for a specific project – including scenarios where the worker is pressured to skip a check, and Fred reacts to how they handle it.

Banking. Employees rehearse their hardest customer conversations – hardship, vulnerability, complaints. With a high-fidelity AI customer trained on hundreds of scenarios and the bank’s own product and compliance knowledge. It reacts to every word, so staff practise judgement, not scripts.

Retail. An AI customer walks up with a problem “I’ve got a leaking tap and no idea what I need” and the team member has to ask the right questions, diagnose the job and recommend a solution, with live AI prompts and feedback.

Industry. In mining, construction and manufacturing, users train on new equipment via a hyper-realistic digital twin with step-by-step instructions, asking questions and completing assessments to build job-ready skills – safely, repeatably and with real-time feedback.

Management skills. Emerging managers rehearse their toughest people moments before facing them for real with an Ai avatar ‘Maya’ who is a team member whose performance is slipping against a deadline, and reacting in the moment to whatever approach they take.

The bottom line: bringing your AI to life

Immersive AI pairs hyper-realistic, engaging characters and environments with an organisation’s own knowledge and context. The result:

• Deeper engagement and genuine participation
• Real behaviour change, not just knowledge recall
• Objective, real-time feedback and benchmarking
• Consistent, repeatable training at scale
• Delivered on headset, desktop, laptop or mobile
• A more capable, safer and more confident workforce

Ready to bring your AI to life???