Regis CIO shares his tips on bringing people along on the AI journey

In September 2025, Regis Aged Care, one of Australia's largest aged care providers, rolled out an AI assistant for its clinical care managers. Built with Microsoft Copilot Studio and Microsoft Foundry, RegiCare Assist is now used by some 150 staff daily, enabling them to spend less time on paperwork and more time with residents in their care.

Regis chief information officer Imtiaz Bhayat shares tips on how he rallied colleagues along this AI journey.

1. Start with a pain point workers already feel

Previously, our clinical care managers would spend the start of their day reading up to 80 pages of progress notes on the well-being of residents in our homes. These notes inform them who needs urgent care and urgent attention, and helps them work out what to do.

We thought: With this large volume of unstructured text, it's really an opportune use case for AI. If you hit the "immediate clinical concerns" button on RegiCare Assist, the AI goes through all the notes and surfaces the urgent issues in the first few minutes rather than in hours.

Healthcare professionals want to be hands-on. This was about making sure their lives are easier and that they can spend more time doing what they love, which is looking after residents. So, anything we could have done to get them back to what they love doing, they jumped on.

2. Be honest about what AI can - and cannot - do

We knew AI was really good at absorbing large blocks of text, picking up sentiment, picking up key themes. And with enough opportunity to fine-tune the AI tool, we knew we could ensure a good level of accuracy.

We were worried about hallucinations and bias though, and we were worried that in a home that's got 150 to 200 residents, whether all the issues would be picked up.

So, we made sure we got as much capability and learning about AI as possible. We talked to the clinicians; we compared our model to what different people said about clinical issues. We took a research-based approach: This is where the model shines and this is where the model could be improved.

All these things helped build a good user experience and a high level of trust. Because they could see we've done all the legwork to assess the strength of the model.

3. Make AI easy to use - and hard to misuse

We made it easy for the users because we spent a lot of time engineering the prompts. The RegiCare Assist screen has nine distinct icons at the top, categorized by common issues like clinical trends, agitation and signs of pain or infection. So, you click one of those icons, and the AI tool provides clear insight into that specific issue based on the information within the progress notes. Those nine icons have got pages of prompts behind each of them. Instead of asking open-ended questions, users can just click the ready-made prompts.

When you're building prompts, you have a structured approach; you think detailed, you think comprehensive, you think about limiting risk. Subtleties around how we developed each prompt ensured that we didn't miss any key details about our residents.

All this resulted in a good user experience. There was a high level of trust with information coming out of the AI.

Read more on how Regis used AI to help its staff focus on residents, not paperwork.