Whether Heidi makes it or not as a global AI clinical workflow layer player, the technology it showed off last night has a clear message for every entrenched medical software player in the country: the moat you thought you had just got a lot narrower.
“Give the computer work back to the computer” is a pretty good tag line.
It sounds obvious but maybe cheesy. After all, computers are at the beating heart of what doctors do every day, right?
Wrong, according to the technology Heidi says it has developed and is developing now, as presented in their Heidi II launch last night.
It’s a pretty good pitch because when you follow their demos – they surely aren’t as smooth in practice but, hey – you can see that the vision is now realisable with AI. And soon.
Health Services Daily has been tracking Heidi for a while.
We called it the Kraken when it announced it was coming for everyone. We then wrote about its strategy problem: a company expanding wide when its key vulnerability was the depth of integration that incumbents like Best Practice had denied it.
It seems like, in the months since, the Kraken was pausing to catch its breath.
It wasn’t. It was loading up.
Heidi II is a claim, demonstrated, not just announced, that the company has found a way to make the integration problem with incumbent EMRs potentially irrelevant over time.
If you are Best Practice, or Lyrebird, or any of the new local AI or cloud players building quietly in Heidi’s shadow, last night’s launch videos should have your full attention.
The skeleton key: RPA and Best Practice
It was hard to decide which was the strategically most critical announcement Heidi made last night: it’s new “magical” agentic workflows – letting loose those pesky agents to do all the boring work in a creative manner – or the confirmation that Heidi’s robotic process automation (RPA) works with Best Practice.
Yep, it’s clunky now, but it’s not so much the current product as the idea it represents.
Heidi is setting about training AI agents to go and get a patient’s record autonomously out of an EMR that doesn’t want to play ball with it and allow it to integrate properly.
That is, I’m a GP working in a Best Practice practice and I love Heidi and Best Practice. Now I can say to my Heidi agent, do me a favour and get my patient list tomorrow out of Best Practice, complete with longitudinal clinical record, and get me organised for all these patients for a good day’s work tomorrow.
No doubt its clunky and slower than a native integration, and that you can’t get it all, and that there will be the odd hiccup and even risky complication.
Heidi would prefer the integration. But if there isn’t one, they are going around it.
It presents quite a dilemma for an incumbent like Best Practice which prides itself on customer centricity – and frankly has been great at that part of their business.
What happens if a hoard of its users fall in love with all the new agentic stuff – see our other article focused on doctors on the same launch – that Heidi has now, and if Heidi is doing it better than Lyrebird, because with all the money they have, they might easily be able to do that at speed.
Do you say to your trusting customers: “Hang on, you can use Lyrebird, our preferred AI partner?” What if Lyrebird hasn’t got the capital to keep up and Heidi is just better? What then? You keep locking Heidi out?
The RPA stuff isn’t so much a great alternative to a deep integration – it’s not – it’s the idea of it: you can if you want to build an agent quickly and just have it do what you’d do if you were a human doing it. Because that is Heidi’s thing. That’s computer work. It will do that for you.
Best Practice controls roughly 80% of the Australian GP software market and has been Heidi’s most significant structural barrier to the GP market so far in Australia.
Its strategic investment in Lyrebird – granting Heidi’s local competitor the deep system access it denied Heidi – looked like it was designed to lock Heidi out of Australian general practice entirely.
Deep integration with the longitudinal patient record is not a nice-to-have for a tool that wants to do anything more than transcribe. It is essential.
RPA changes things. A GP who loves Best Practice and also wants everything Heidi II is now offering has, according to Heidi, a path now to build a practice environment where they use both.
If enough GPs start running Best Practice and Heidi in parallel, that will eventually create real commercial pressure on Best Practice to give its customers the deep integration they are already building workarounds to obtain.
Heidi is betting that once doctors have decided which tool they prefer, the market will have to follow. It is a fascinating long game.
Related
HealthLink integration
Beyond the RPA story, one integration confirmed at the demo deserves an honourable mention. Heidi has connected to the HealthLink directory – Australia’s dominant secure clinical messaging network.
A GP can now dictate a referral letter and send it as a secure message to any specialist or service in the HealthLink directory, directly from within Heidi, at a voice prompt.
An audience member asked the question plainly at the demo: “So I could send my letter straight from Heidi rather than separately using HealthLink?”
The answer was yes.
Hospitals, pricing, and the funding question
The hospital play of Heidi II is less developed but no less ambitious.
Heidi is about to be embedded in Metro South Queensland (not connected to the EMR though), it is close to a formal pilot with Eastern Health in Melbourne – a big urban network without a full EMR, which makes it a particularly attractive beachhead.
It is almost certainly among the NSW Health panel scribes which still haven’t been announced.
Against Epic and Oracle, Heidi’s advantage is agility in innovation, speed to market and the fact that hospital doctors have never particularly liked the systems their administrators bought.
Pricing is still a big unknown for all of the Heidi II stuff as it’s agentic and will be token based.
It’s all going to use a lot more compute than the scribe, but just how much is anyone’s guess. It will be dependent a lot on what agentic functions an institution wants in its deployment.
The first month of agentic access is free for clinician-tier subscribers; a credit-based add-on follows. Heidi’s stated position is that they will absorb compute costs while they work out the economics.
Is that a nice gesture or entirely strategic?
Essentially Heidi is going to pay for all your doctors to get addicted to it for free. But it’s not going to be free.
I think I read about this tactic in a great book called Narconomics. Get the users addicted to supply then start talking about price.
Agentic AI uses substantially more compute than a scribe. Someone will pay for it eventually, and the moment Heidi moves from absorbing that cost to passing it on will be a significant inflection point – both for Heidi’s unit economics and for the competitive calculus doctors are running.
As impressive as Heidi II is, speed, innovation and good marketing aren’t all the group is going to need to make it.
It’s reasonably telling that they didn’t close this round with any new US investors.
The official framing: too much to invest in in AI right now. That may be true. It is also the kind of statement you make when the alternative explanation is less comfortable. Heidi is competing globally against US and UK competitors with far larger capital bases and in the eyes of the big money betting on medical AI, there are better bets.
The feature set announced last night is impressive. But with enough money and a well led group, it’s easily replicable.
For Lyrebird, Myscribe, MBSpro and the other local scribes, this launch is a material escalation of local competition.
No Australian competitor has assembled anything close to this feature set publicly. For Best Practice, the threat is no longer theoretical: its customers now have a credible path to running Heidi alongside it, and enough of them doing that should eventually force a reckoning.
Whether Heidi makes it globally is uncertain.
Whether the technology it showed last night will transform how Australian doctors work – regardless of who ultimately delivers it – is not. That transformation is coming. Last night was a significant bet that Heidi will lead it.
What that looks like from inside a consulting room is a separate question – and in some ways the more important one. See here for that take.
The Kraken is back … for now anyway.