AI is suddenly challenging the premise that software as a service healthcare platforms are highly defensible (and therefore valuable) long-term business models. There are big implications for the medical software sector, provider productivity and, ultimately, patient outcomes.
If you’re a medical software business CEO with an eye on any of the global technology indexes in the last six months you’d be silly not to be seriously paranoid about what AI is doing to valuations of once seemingly rock solid software platform businesses.
Xero is down 62%, Atlassian 58%, Pro-Medicus (a medical imaging data platform) 63% and local logistics platform Wisetech is down 62%.
Notably, most of these groups are not declining in revenue or profitability. The drop in valuation is driven mainly by the market making predictions of how AI is starting to undermine some of the core defensible properties of these groups, mainly in how easy and inexpensive AI is starting to make software development.
How might this so called “SaaSpocalypse” unfold for our most important medical software platforms, both local ones like Best Practice and HotDoc, and even some of the globals embedded in our state-based hospital systems, like Epic and Oracle (Cerner)?
This article originally ran on TMR’s sister site, Health Services Daily. TMR readers can sign up for a discounted subscription.
And what might this disruption offer, or not, to our providers – in the way of productivity – and patients, in terms of outcomes, and in what timeframe?
That’s a lot of whats.
Why there is paranoia at the top of some of our better medical platform businesses is that mapping any set of likely outcomes at this point of time to this very serious disruption is near impossible.
The only thing that is clear is that none of the existing businesses can sit on their long-term market moats, like some of them have done for quite a few years now and ride this thing out.
Every major software platform business we have has to be looking very carefully at their market and client dynamics, and the opportunity those dynamics present to rapidly embed meaningful AI into their functionality for their existing client base.
Overseas it’s simplistically being termed as a need to move from being SaaS to intelligence as a service (LaaS). That is, take your current business of connected screens to do process, and use the client base and access to deliver solutions and intelligence using new AI agents.
The key for most of our important healthcare SaaS platforms is how quickly and intelligently they can deliver enough of this intelligence to their existing client base to stave off smart, well-funded challengers.
The money, scale, existing market power, and deliberately established information and data moats that can make leaving a software platform and trying to transition to a new supplier near impossible, exist in spades for our big global EMRs.
Epic is moving particularly fast with AI, moving well beyond the basics of productivity in early AI tools like scribing and note summaries, to deeply embedded AI-driven decision support integrations with key clinical decision support providers such as the USD$12 billion funded OpenEvidence from the Sutter group.
Epic is embracing the idea that by partnering with patient-facing AI agents they will be one day able to deliver decision support with full context of a patient’s clinical data embedded in the support.
That’s not entirely a pipe dream if you consider how fast patients are turning to health-specific AI agents such as ChatGPT Health as their partner in their healthcare journey and loading them up with everything they can get in terms of healthcare data on themselves.
Epic is also being strategically clear: it can’t be a clinical decision support data group, or an AI specific healthcare engine, but it can deliver both in the context of what they already do across thousands of hospitals.
A key for Epic is it has the money and infrastructure to move fast to secure its previously giant data moats. It’s obviously also got some clever leadership.
Related
Where do our key Australian software platform vendors get that money and existing development infrastructure though, particularly if, like a few of our big patient management systems, they’ve been sitting on their monopoly and data moat protection in relative terms for quite a while.
You can point to any of our important and large healthcare SaaS platforms – Best Practice, MedicalDirector, HotDoc, Gentu, Healthengine, Alcidion, maybe, and even Global Health – and they all have a problem: they don’t have a lot of capital to invest and move rapidly.
They are all still using legacy software platforms which often still involve patchwork connections from the cloud to what essentially are still desktop local server-based platform architectures (not Alcidion), and they all have had big protective data and UI moats that have protected them from smart cloud and AI-based competition.
By the way, all these businesses have very large and tightly held clients, with moats that are going to keep them in play for a while yet, even if they don’t move quickly.
But scroll through the entire membership list of the Medical Software Industry of Australia and you are almost certainly reading the names of a lot of businesses that aren’t going to make it through this disruption.
There are a lot of businesses with single point-to-point software solutions, which, while nifty for solving one or a couple of specific problems today, are going to be murdered quickly, either by well-funded new competitors, building their software for a dime, but adding a lot of additional intelligence around broader functionality, or, by the existing big players – local and global.
Agentic AI agents, a big sticky customer base and more money are going to be almost impossible to compete with.
If not Best Practice, then a well-placed cloud competitor or start-up, or indeed an Oracle or Epic are likely to be adding a lot of functionality quickly at their front ends to rapidly attempt to create more customer stickiness.
One problem for Australia’s big local groups might be the likes of AI groups, like Heidi, that start getting momentum on both accessing their customer base, and through that, broadening functionality rapidly through the much faster and cheaper AI software development protocols we are seeing with AI.
A key differentiator for survival here might be the size and complexity of your existing client base.
In this respect B2B healthcare software platforms – Best Practice, Gentu, Orion Health, and a few others – are far more exposed than the consumer-facing platforms like HotDoc and Healthengine.
Heidi is hiring most of anyone who is anyone in digital health in Australia because it can. It is already building a huge overlap of customers that use some of our major local platforms like Best Practice.
It’s almost certain that Best Practice decided earlier this week to release a free version of Lyrebird in an attempt to slow down how fast Heidi is acquiring their clients, even though at this point of time Heidi isn’t acquiring them with a view to competing with Best Practice as a patient management system.
But if Heidi is currently funded to $100 million, and keeps gaining momentum, and sees a need in its global aspirations to add a lot more functionality to its customer set than the simple stuff it has started with – which it must do – how long before it starts competing far more directly with some of our local B2B platforms?
Best Practice is sitting close to Sonic, but not owned by Sonic yet. It hasn’t got much current access to capital. It’s not its DNA or its style and as far as we know it’s not talking to anyone about building a war chest of money and people to defend what most people see as one of the biggest healthcare platform monopolies in the country.
(Note: I’m not criticising Best Practice for being in this position or getting to it. Quite the contrary, this has been a very clever company over the years, and if you live in Australia as a software platform business – most bigger businesses – monopoly is the play.)
One of the biggest problems for our incumbent B2B platforms is going to be keeping up with the functionality their doctors may need to be able to handle the massive disruption all our doctors are about to experience with their patients through the patient-side AI agents.
The huge asymmetric information imbalance that has existed with patients and doctors is rapidly disappearing.
Will our patient management systems need to address this problem for their doctors quickly? How?
Epic is doing it by partnering with and incorporating a consumer AI agent directly into its EMR, a move that promises to not only bring superior clinical decision support functionality, but perhaps even embedded individual patient data to work with that functionality.
Can Best Practice, with a leadership culture of “do it slow and steady to get it right”, best in class, and customer first, but very little money, stave off a Heidi, if Heidi sees the need to come for it.
What if Heidi starts embedding clinical decision support and does a deal with patient-side AI?
Epic, of course, is already going in this direction and we know it also has a well-developed ambulatory care (primary care) module in the US? How might that go if it all works on the cloud and can talk to all the hospitals in NSW, Tasmania and a few big and important ones in Victoria?
If someone could crack open Best Practice, then you’d have to think that all bets would be off for most incumbents, especially those still sitting on old tech stacks, and that’s the majority of our members of the MSIA.
What about the consumer-facing businesses like HotDoc and Healthengine?
With so much extra data in the form of patient data they may have a little more time to adapt. But they will need to adapt, and quickly.
This all means new rounds of investment.
Is that what HotDoc’s new private equity owners had in mind when they paid 27 times EBITDA for the group a few weeks back?
If ChatGPT Health and Claude Health are as good as people are saying already, then how long will it be before these giants become a default digital front door for most Australian patients?
The federal government and Healthdirect are very smartly already contemplating partnering with these AI agents and letting them have the national digital front door in exchange for inserting themselves into the AI process on behalf of patients for when a real doctor, or pharmacist does actually need to be handed off by the AI.
Maybe the lock-in that HotDoc has with app downloads on so many patients isn’t anywhere near enough.
After all, how long before it has most of HotDoc’s patient data and decides to create its own booking functionality, perhaps free to all GP practices?
OK, I’m stopping here.
There’s a lot of whats, ifs and buts here, as you can see. And plenty of reason for a software platform vendor CEO to be very paranoid.
