Microsoft has disclosed that Regis Aged Care, one of Australia’s largest residential care providers, deployed a custom AI assistant called RegiCare Assist in September 2025. The tool, built with Microsoft Copilot Studio and Microsoft Foundry, is now in use across 72 care facilities, with approximately 150 staff members leveraging it to streamline clinical documentation. Its primary function: summarizing nursing notes while keeping a clinician firmly in the loop for final approval. This human-governance model signals a cautious but pragmatic approach to bringing generative AI into highly regulated aged-care workflows.

The announcement landed through Microsoft’s enterprise channels, spotlighting a real-world implementation where AI shoulders the grunt work of distilling handover notes, progress updates, and care observations. But unlike a black-box automation, RegiCare Assist explicitly bakes in a human review step. For any summary it produces, a registered nurse must validate and sign off before the content enters the resident’s official record. This design is not incidental—it is central to meeting Australia’s Aged Care Quality Standards and the broader clinical governance frameworks that govern aged care.

The Documentation Crunch in Aged Care

Care minutes are precious. Yet nurses and personal care workers in residential facilities routinely spend up to two hours per shift writing notes—converting fragmentary jottings, verbal handovers, and observed changes into structured records. Those notes must be accurate, legally defensible, and compliant with funding documentation requirements. The cognitive burden is immense, and it eats into time that could be spent one-on-one with residents.

Regis Aged Care, which operates dozens of homes across Australia, sees this tension daily. Its workforce, like that of the broader sector, faces chronic shortages and burnout risks. By offloading the initial drafting of summaries to AI, the organization aims to reclaim nurse time without sacrificing the rigor of clinical records. The early data point—150 active users across 72 sites—suggests a phased but organization-wide rollout, not just a pilot tucked in a single innovation lab.

What RegiCare Assist Actually Does

Based on the limited technical details Microsoft has shared, RegiCare Assist ingests raw nursing notes—likely a mix of structured template fields and free-text narratives—and generates a concise, coherent summary. The output is not final; it is a draft that a human clinician reviews, edits if necessary, and then approves. The assistant’s interface is built with Copilot Studio, Microsoft’s low-code platform for creating custom AI copilots. On the back end, Microsoft Foundry orchestrates the AI calls, grounding responses in trusted data and likely applying retrieval-augmented generation to pull in facility-specific policies or resident history where relevant.

Microsoft Foundry, formerly known as Azure AI Studio, provides the governance scaffolding: content safety filters, prompt engineering controls, and integration with Azure OpenAI Service models. This architecture means RegiCare Assist can operate securely within Regis’s compliance boundary, ensuring that no sensitive health data leaves its controlled environment. The “human governance” is not just a workflow step; it is enforced by the application design—the system will not finalize a note until a credentialed user explicitly accepts the AI’s suggestion.

Why Human Governance Matters Here

Aged care is not a low-stakes domain. A poorly summarized note could obscure a critical change in a resident’s condition—missed signs of a urinary tract infection, an undocumented fall risk, an omitted medication reaction. Regulators, families, and auditors scrutinize these records. So the approach Regis and Microsoft took is instructive: treat AI as a capable but fallible scribe, not as an autonomous clinician.

This human-in-the-loop model is becoming the standard for clinical AI assistants. Ambient listening tools like Nuance DAX (also a Microsoft product) similarly transcribe and summarize patient encounters but require the physician to verify the final note. RegiCare Assist extends that pattern to aged-care documentation, where the interaction is often asynchronous—notes dictated at the end of a shift, not captured in real time during a consultation. The governance layer ensures that every AI-generated summary carries the accountability of a licensed professional.

Deployment Scale and Early Signals

Regis’s rollout across 72 facilities is not trivial. It suggests the organization has invested in change management, staff training, and IT integration. Embedding AI into an existing electronic health record (EHR) workflow is notoriously sticky; aged-care systems are often fragmented, mixing modern cloud platforms with legacy on-premise databases. That the assistant is in daily use by 150 staff implies more than a technical success—it points to user acceptance, likely because it saves documentation time without threatening professional judgment.

One nuance: 150 users across 72 facilities means roughly two users per site on average. This could indicate that usage is concentrated among senior nurses or care coordinators who generate the most complex notes, or it may reflect a deliberate, incremental adoption strategy that avoids overwhelming the workforce. Either way, the figure sets a benchmark for similar deployments in aged care.

The Microsoft Tech Stack

Copilot Studio is the star of the show for building the conversational experience. It allows non-developers—or “fusion teams” combining clinicians and IT—to create custom copilots that plug into internal data sources. In this case, RegiCare Assist likely presents as a pane within the EHR or a standalone web app where a nurse pastes notes and receives a summary. The underlying prompt engineering would need to be carefully tuned for clinical language: it must understand shorthand, abbreviations, and domain-specific terms (e.g., “ADL,” “CVA,” “NFR”) without hallucinating or omitting critical observations.

Microsoft Foundry ties together the AI safety features. It provides model selection (likely GPT-4o or a healthcare-tuned model), grounding in enterprise data, and monitoring for responsible AI outputs. For an aged-care provider, Foundry’s ability to enforce privacy is non-negotiable. Regis can also track metrics—summaries accepted without changes vs. those heavily edited—to refine the assistant over time.

Broader Industry Implications

RegiCare Assist is part of a wave of “AI scribe” technology moving from physician offices into long-term care. In Australia, the Royal Commission into Aged Care Quality and Safety (2018–2021) put a harsh spotlight on documentation failures. Any tool that improves the accuracy and timeliness of records while reducing staff burnout aligns with regulatory momentum toward transparency and quality improvement.

Microsoft’s positioning here is strategic. By showcasing a custom Copilot Studio application in aged care—a sector often underserved by cutting-edge IT—it demonstrates that its AI stack can adapt to niche, high-stakes environments without requiring massive in-house AI engineering teams. The combination of low-code development and robust governance may encourage other aged-care providers, not just in Australia but globally, to explore similar solutions.

Potential Pitfalls and Unanswered Questions

The announcement leaves several gaps. We don’t know the baseline accuracy of the summaries or the rate at which nurses must intervene. Clinical language is messy; AI can misinterpret severity or omit relevant context. The fact that Regis adopted a human-review step strongly suggests that the system is not yet trusted to run autonomously—which is prudent but also hints at current limitations.

Integration with existing systems is another unknown. Does RegiCare Assist sit natively within Regis’s EHR, or does it require copy-paste between applications? The latter would dampen the time savings and increase the chance of error. Also, data residency and sovereignty are critical in Australian healthcare. Microsoft clarified that data remains within Regis’s tenant, but without auditing the specifics, we can only take that at face value.

Finally, the “150 staff users” metric is vague. Are those daily active users? Total trained? How many hours per week does the tool save? Without those numbers, it’s hard to gauge return on investment. Still, an aged-care organization moving this quickly—just months after Copilot Studio became widely available—signals a strong belief in the approach.

The Human Element Endures

For Windows and Microsoft enterprise watchers, RegiCare Assist is a compelling case study. It’s not about flashy demos but about grinding out real utility in a sector where the consequences of AI error are measured in human wellbeing. The design choice to keep a nurse as the final arbiter is not a weakness; it’s the feature that makes regulatory approval plausible.

As AI documentation tools proliferate, the Regis example reaffirms that governance is not a bolt-on but the architecture itself. The technology stack—Copilot Studio for the front-end experience, Foundry for safe AI orchestration—shows that Microsoft is building with this reality in mind. Whether RegiCare Assist scales beyond note summarization into medication reconciliation, care planning prompts, or family communication drafts remains to be seen. But the foundation is now field-tested across 72 care homes, with nurses, not just engineers, shaping the outcome.

For the aged-care sector, this could be the blueprint. For Microsoft, it’s validation that Copilot Studio can tackle enterprise-grade, compliance-heavy scenarios. And for the 150 nurses already using it, it’s one less hour spent typing—and one more hour spent caring.