There is a large and largely overlooked segment of Australian hospitals that has no enterprise EMR, no plans to get one, and a clinical workforce quietly using whatever AI tools it can find. Heidi II’s agentic capabilities potentially suit a tilt at this market now.
The Australian hospital software market is often discussed as though it is dominated by Epic and Oracle Cerner – two American enterprise giants with deep roots in the major metropolitan health networks of New South Wales, Victoria and Queensland.
That framing misses most of the country.
Tasmania’s entire public hospital system has no enterprise EMR. It runs Infomedix – a digital patient chart and clinical record viewer, not a full transactional EMR – which means patient data is captured, digitised and indexed, but the structured workflow automation and clinical decision support of a true EMR simply do not exist.
Western Australia’s public system is a patchwork waiting for the money and will of that government to move on an EMR solution. South Australia and the Northern Territory are thin on enterprise coverage outside major centres.
The private hospital sector is complex.
St Vincent’s Health Australia signed a contract to roll out MEDITECH Expanse across its 10 private hospitals in Victoria, New South Wales and Queensland – a genuine enterprise EMR deployment, though still in progress.
But St Vincent’s public hospitals in Sydney and Melbourne are a different story: a $62.5 million business case for a full enterprise EMR was knocked back by NSW Health for state capital funding, and those facilities run Infomedix instead, the same digital chart system used across Tasmania.
Ramsay Health Care – the largest private hospital operator in the country – partnered with Infomedix and TIMG to roll out a national digital medical record solution across its hospitals.
St John of God uses Infomedix specifically for its coding manager function, not as a hospital-wide clinical system.
Across every state, a very substantial portion of Australian hospital bed-days are managed in environments that have no enterprise EMR, only a digital chart, a departmental clinical information system, or paper-based workflows that have been digitised but not integrated.
The screen reader question
Infomedix describes its own platform as a clinical information viewer rather than a transactional EMR.
It scans, digitises and indexes physical clinical workflows and makes them accessible via a cloud-hosted digital chart. Clinicians can access complete patient histories quickly. What they cannot do is the kind of structured write-back, order entry, decision support and cross-system workflow automation that a full EMR enables.
Altera Opal is a similar product and is deployed across a lot of WA.
Heidi’s RPA approach – agents navigating the screen as a human would – is, in theory, applicable here, just as it is for Best Practice in general practice.
If the record is visible on screen and the clinician is logged in, an agent can read it.
In practice, the value proposition is narrower than it would be against a full EMR. Infomedix and Altera Opal hold appointments, clinical notes and patient history – enough for Heidi to pre-chart a day, draft a discharge summary, or flag a follow-up.
For scribing and knowledge retrieval, Infomedix is essentially irrelevant to Heidi’s core loop: the agent works around it, not with it. For anything requiring structured write-back to the record, Infomedix was never going to provide that regardless of what AI tool sits above it.
Related
The shadow IT reality
Across the non-EMR hospital sector – and, quietly, within many EMR hospitals now as well – Heidi has a significant shadow IT presence.
Clinicians using Heidi for scribing and knowledge retrieval inside hospitals that have not sanctioned it is not a new phenomenon.
Heidi II adds a reason for hospital leadership to pay more attention to what might now be possible: agents that can do work across a clinical day, not just transcribe, summarise and provide advice on a consult.
For a hospital without an enterprise EMR, the calculus is different from an Epic or Oracle site.
There is little or no incumbent vendor relationship to protect, no sunk-cost integration to defend, and no CIO whose career is tied to a platform decision made five years ago.
The question is simply: what does this do for my clinicians and patients, what does it cost and therefore what is the ROI?
That ROI question has always been a lot more important for the private hospital system than the public one.
What does Heidi CEO and founder Dr Thomas Kelly think?
“In those enterprise settings, computers – for now – I don’t think a CIO is going to underwrite the computer use part of Heidi II. But the agents bit – we’ve had lots of interest for outpatients and coding of visits.”
Dr Kelly’s distinction matters.
The RPA agents – the ones that navigate patient records autonomously under a clinician’s login – are the part that makes hospital administrators nervous, for understandable governance and liability reasons.
The agentic features that don’t require computer use are a different conversation: discharge summary drafting, outpatient coding assistance, referral letter generation, evidence retrieval for a ward round. None of those require Heidi to take over a screen. They require access to data and a clinician’s instruction.
In a hospital with a Microsoft data lake, or with structured data landing anywhere accessible, those workflows are available now, regardless of whether the underlying system is an enterprise EMR or an Infomedix digital chart.
AI is making some hospitals rethink EMR investment altogether
There is a quieter conversation happening in Australian hospital leadership that is also directly relevant to Heidi II (and anyone trying to catch it).
Some health services that have not yet committed to a full enterprise EMR are now pausing that decision. The logic of a $200 million, decade-long implementation has shifted under the weight of what AI is now demonstrably able to do.
St Vincent’s public hospitals are an instructive case.
The NSW Health funding knockback was not because the business case was weak: capital is scarce and a full enterprise EMR is enormously expensive to implement and run.
It’s existing environment – digitised records, no full EMR, clinicians wanting better tools – is where an AI clinical layer running over an Infomedix chart might just close a meaningful productivity gap.
This isn’t an AI replacing an EMR.
It is that the gap between a well-resourced AI clinical layer running over imperfect data and a fully implemented enterprise EMR is narrowing faster than anyone expected.
And the cost differential might be enormous – everyone still has to see what compute costs when agentic is deployed in certain situations.
But for a health service without an existing EMR, the question is no longer only “can we afford an EMR and if so which one?”
Now, it’s “should we buy one at all, or invest in AI infrastructure and integration to existing systems instead?”
Mostly none of this will apply to health services already deep into Epic or Oracle implementations. For them, the sunk cost and compliance dependencies make a course reversal fiscally implausible.
But for Ramsay, for Tasmania’s public system, for St Vincent’s public hospitals, for every mid-market country hospital running a departmental clinical information system that was never designed for this era, the question is now live.
Heidi is not the only beneficiary of this shift. But it is the most visible one making the case publicly.
Dr Kelly’s hospital strategy is not to defeat Epic or Oracle on their own turf. It is to win the hearts and minds of clinicians strongly enough that integration becomes inevitable, and maybe to take as much of the non-EMR market as possible in the meantime.
The full story of what Heidi’s play now looks like inside major enterprise EMR hospitals Epic, Oracle, and the shadow IT dynamic that is making their CIOs uncomfortable is a separate and equally significant story. It follows next week.
The potential and emerging place of agentic AI and AI scribes in the Australian hospital market is the main theme of our upcoming AI and new models of care reshaping hospitals summit in Melbourne on 29 October. You can still get a ticket HERE. Use this discount code to get 20% off all ticket types: FUTURE20