Tend’s Josh Robb: AI, Cybersecurity, and Healthcare Transformation

Transcript is computer-generated and may contain errors.

Paul Spain:
Greetings and welcome along to the New Zealand Tech Podcast. I’m your host, Paul Spain. Joining us today is Josh Robb, co-founder and chief operating officer at Tend Health. Josh has spent more than 2 decades building and scaling technology teams and products with leadership roles across Pushpay, AIA, and Tend. Tend is reimagining healthcare for New Zealanders, combining a nationwide network of clinics with 24/7 online care and a technology platform designed to make accessing healthcare simpler, faster, and more connected. Josh, welcome back to the show.

Josh Robb:
It’s great to be here, Paul.

Paul Spain:
Good to have you here. Lots to delve into today. So looking forward to hearing an update on what’s happening in the Tend world, but we’ve got the usual sort of local and international tech news to delve into too. And of course, a big thank you to our show partners. So thank you, Spark, One New Zealand, 2degrees, Workday, Fortinet, PwC, and Gorilla Technology. On the New Zealand front, New Zealand is expanding our use of maritime drones for surveillance as we see allies including the United States and Australia increasingly deploying armed sea-based drones in real-world military operations. And that’s really sparked, I think, a bit of interest and maybe some, you know, concern. Should New Zealand be focused on just monitoring and using, you know, drones from a surveillance perspective across the oceans? Or should we be adopting drone capabilities that can, shall we say—

Josh Robb:
Defences.

Paul Spain:
Yeah, take action, take offensive action. And we’ve, you know, we spoke earlier on in the year with Scios Aerospace and heard a little bit about what they were doing. They’ve had huge success in the UK with the British Army. And, you know, we now know that the New Zealand Defence Force has moved ahead and has purchased SIOS SM300 ocean-going drones. However, yeah, at this stage, more from the surveillance perspective.

Josh Robb:
Yeah.

Paul Spain:
What are your thoughts on this? Because there’s the angle that, you know, if we put people out there to take action, then that puts people’s lives at risk. However, if we go fully autonomous and you’ve got AI and automation— Like, what the hell? Making decisions, then that’s a problem too.

Josh Robb:
Yeah, I mean, it’s tricky, isn’t it? I think New Zealand has the largest exclusive economic zone in the world around us to patrol and a very underinvested military to do it. So some assistance seems like a no-brainer. But I mean, I don’t think the P-3s that are flying around doing the aerial surveillance are armed. So do we need armed drones? What are we going to do, attack Chinese fishing fleets that are kind of— it seems, I don’t know. Yeah, it doesn’t seem very Kiwi, does it? But at the same time, I think it’s fascinating watching what’s happening, what Ukraine are doing with the Black Sea Fleet. They’ve taken out so many tankers in the last couple of weeks. I don’t know if you’ve seen any of this coverage. But I mean, they’re just absolutely destroying the Russian kind of black, what do they call it, shadow fleet, whatever it is, that are kind of exporting illegal oil.

Josh Robb:
And I think they’ve taken out 30 or 40 vessels in like a week, I saw last week or the week before. So it’s a pretty important capability to be on top of. And if we’re not patrolling our EEZ with armed drones, someone else will be. Like, I don’t know, it seems like a—

Paul Spain:
Yeah.

Josh Robb:
Yeah.

Paul Spain:
And this is that whole balancing act of getting the best, the positive, good uses out of technology and the reality that, you know, as a country, we have to have a defence capability. And when it comes down to being cost-effective, leveraging newer, smarter technologies—

Josh Robb:
It seems way smarter to invest in that than in new ships, right? New personnel ships, new cosmocong ships.

Paul Spain:
I think that’s the cost of cost differential. between traditional military investment, whether it’s ships, jets, helicopters, is night and day different.

Josh Robb:
Yeah.

Paul Spain:
So there’s a degree there where I’d be leaning in a lot more towards what CSIOS is producing and what can be done on that sort of basis that puts people less in harm’s way. I think there is though a bigger picture beyond just patrolling, you know, our own sea area because New Zealand Defence Force—

Josh Robb:
What other capabilities do we need?

Paul Spain:
You know, has routinely, you know, served in other parts of the world. If we, you know, if we look back over time, they’re not just exclusively here. So having that capability and being able to use them appropriately in the future, I think, is something that we definitely have to keep exploring. And, you know, part of that exploring means investing in actually, you know, getting that capability on board.

Josh Robb:
But probably also interoperability with partners.

Paul Spain:
Yes, I think there’s a lot of military clout that, you know, we see other countries having that New Zealand is unlikely to have any time soon.

Josh Robb:
Yeah.

Paul Spain:
Whether that’s nuclear subs or what have you, right?

Josh Robb:
Planes, whatever. Yeah, yeah.

Paul Spain:
So on a, on a, some, yeah, very much an international story, but with implications for New Zealand, we hear that Indian startup Skyroot Aerospace has successfully launched their Vikram-1, which is India’s first privately developed orbital rocket. Now, this milestone supports India’s ambitions to expand what they’re doing in the space sector, particularly from a commercial standpoint, as they work to really be able to compete strongly on a global basis when it comes to, you know, satellite launch and other private space sector innovation. So yeah, exciting news for India. But interestingly for us in New Zealand, the Vikram-1 rocket Actually, I guess you could call it something of a potential competition for what Rocket Lab are doing. What Rocket Lab are doing, yeah. 350 kilograms into low Earth orbit is what the Vikram-1 is able to lift, and that compares to around 300 kilograms that the Electron launches routinely from New Zealand on a pretty regular basis. Now, of course, if we step back a little bit, there are, there are a lot of companies that have had a success at being able to get, you know, 1 or 2 rockets launched, but very, very few who are able to do anything on a regular—

Josh Robb:
Reusable.

Paul Spain:
You know, cadence, right? So, so yeah, really interesting to see. I see there that India are aiming to grow their space economy from what’s listed as $8 billion to $44 billion by 2033. So look, they’d be silly not to. I think, you know, any country that has some involvement in the space sector should be smart enough to look at, hey, how can we How can we partner this?

Josh Robb:
How can we get in on this?

Paul Spain:
Yeah, yeah, absolutely. You know, just as New Zealand is with our, you know, roughly 300 aerospace companies and, you know, multiple players, you know, doing some pretty cutting-edge stuff, you know, up and down the country. So yeah, but, you know, when we hear these bits of news, it, you know, in some ways there’ll be, you know, especially Indian Kiwis kind of probably celebrating, think this is really cool. But, you know, there’s also a lot of Kiwis I think the last thing I heard was it’s at least 60,000 Kiwis with shares in Rocket Lab.

Josh Robb:
Yeah, wow.

Paul Spain:
So yeah.

Josh Robb:
I mean, I think the private space race is actually kind of one of the coolest things that’s happened in my lifetime, really. Like, as a kid, you know, you kind of followed what NASA were doing and, you know, sending a rocket up was like a real milestone event.

Paul Spain:
Yes.

Josh Robb:
And now it’s happening daily. And there’s, I mean, I just don’t even think we really yet understand, like we need a lot more lift than we’ve got. So I’m kind of here for competition. It’s obvious that there’s kind of sovereignty desires around not being dependent on a single supplier or a single country or a single geopolitical bloc. So it’s natural. I mean, you would’ve seen the Chinese landing last week, which is pretty cool.

Paul Spain:
Japan are at it.

Josh Robb:
Yeah, yeah, Japan are at it. I mean, it’s just inevitable, you know? So I think it’s great. I mean, I think Russia are really, really hurt by Starlink not making that available to them or turning it off for good reasons. But that kind of like, oh, hang on a minute, this is starting to impact kind of our strategy, our military strategy is kind of a big wake-up call for everyone. Like everyone had to get their own GPS constellations up 20 or 30 years ago. It’s kind of another round of that really of like, oh, hang on a minute. Yeah. So, and I mean, it’ll be interesting to see what happens with the data centre satellite space stuff, you know, if that really turns into a massive pathway, which obviously Elon’s pretty focused on, then I think that will rapidly change the demand.

Paul Spain:
Yeah, there’s a kind of big if still in my head as to whether these things will be pulled off or not. And then if, then what does it actually look like? And how much will we see, you know, that that evolve in terms of data centres being in space versus not. So yeah, lots of—

Josh Robb:
Fascinating times.

Paul Spain:
Lots of interesting times ahead. Now on that front, of course SpaceX and their share price has been fairly constantly getting attention because it was such a big high-profile initial public offering there. And it attracted, you know, a good chunk of investment. We saw the share price, you know, rise very quickly. Now it’s somewhat sort of, you know, come back to earth, as it were. And I read 23,000 New Zealand investors have bought shares through Sharesies, which—

Josh Robb:
Wow.

Paul Spain:
to me, is a little bit of a reflection of the excitement and the hype of new things, new opportunities. And the reality that folks will get involved in something that I would’ve thought was at the more risky end of an investment because this is not a—

Josh Robb:
Risky end.

Paul Spain:
a new entity. They’ve been around for a long time and, you know, they’ve already attracted a really, really large amount of capital. So, you know, it’s been quite fascinating to watch. And then, you know, I see some coverage kind of looking at, you know, the fact that so many other folks have ended up, you know, ultimately with an investment in SpaceX, you know, because they’re part of an index fund.

Josh Robb:
Yeah.

Paul Spain:
And yeah, that I guess for me has triggered some thoughts around, hmm, how, you know, how should we invest in 2026 and how’s that maybe different from, you know, from previous times to invest?

Josh Robb:
I mean, I think, I mean, we’re not financial advisors, right? So there’s nothing here as financial advice, but—

Paul Spain:
This is just us spitballing.

Josh Robb:
Just us spitballing. But I think like I was listening to someone on the weekend talking about this and they were saying you can now get ETFs that are like the top 100 less Elon or the top 100 less whoever you morally object to.

Paul Spain:
Yeah.

Josh Robb:
If you’re a Palestinian rights person, you can get the, okay, we’re gonna take out the cohort of companies that we don’t like ’cause they’re supporting Israel or whatever it is. So you can kind of package products now. And I think if you’re concerned about that, then there are products in the marketplaces that you can buy that kind of give you broad coverage without having specific investments if that’s what you want. The other side of the coin is, I think this is slightly misunderstood, is that indexes are scaled to the free float. So it’s not the market cap and it’s not the valuation. And people kind of lose track of that. It’s like, I think SpaceX is, I think like only 3% free float or something at the moment. Everything else is locked up.

Josh Robb:
Or maybe it’s slightly more than that. But so the actual proportion of SpaceX in your index is not the market cap of SpaceX against the market cap of everything else. It’s the proportion of the free float. So it’s actually a lot less influential than you might otherwise expect. I think the number 3% is actually how much SpaceX— affects the top 100 index. So it’s not actually as influential as its market cap might lead you to think. And so the swings in it don’t necessarily affect your ETF there to quite the same amount.

Paul Spain:
Yeah.

Josh Robb:
But I mean, it is a reasonable question. Like, you know, and I think people are particularly alive to this because of the index rules waiving for the IPO. So like it’s coming. And so you kind of have to worry about, well, is this gonna affect my retirement savings plan or is this gonna affect my, you know, some other scheme that I’ve got going on.

Paul Spain:
And look, I think we’ve got more flexible investment options today than we’ve ever had before, right?

Josh Robb:
Absolutely, yeah.

Paul Spain:
Whether that’s for your KiwiSaver and you want to build a custom KiwiSaver that fits your interests. And if you’re a tech person, you can lean in onto the tech stocks that you’re interested in and get that, you know, custom things. Get exposure to the big ones, yeah. Get exposure to startups. There’s, I mean, it’s actually, you know, I think an incredible time from, you know, from that perspective. And because such a high percentage of the New Zealand population now has a KiwiSaver, then you’ve got a very high percentage of the New Zealand population that are investing outside of their, you know, their home.

Josh Robb:
Property.

Paul Spain:
If they have property, home investment and outside of bank account. So yeah, there’s some really good stuff about that. I think there are challenges in that the level of knowledge and education probably varies quite broadly.

Josh Robb:
Yep.

Paul Spain:
And some of the folks that I chat to, hearing how they talk about what to invest in, is, you know, I find it a little bit head-scratching, but we probably all succumb to a degree to the kind of the hype and the excitement around what this company and that company could or should or might do. And so, yeah, it is fascinating. But, you know, certainly, you know, where I’ve had friends that are like, oh, Paul, should I buy XYZ stock? And I’m not usually one to recommend, you know, stocks at all, even to you know, close friends, I’m usually saying, well, you know, think about these things and recognise that probably—

Josh Robb:
What are the trade-offs?

Paul Spain:
Anything that you know about, whether it’s SpaceX or any other company, that’s probably public knowledge that everybody else has. So it’s quite possibly built into the price already. So, you know, you have to then be trying to weigh up what you know about the risky future. And yes, if they pull that off, it could However many times up. If they don’t pull that off, then, you know, it could go in the other direction. And really, that’s, yeah, that’s part of the reality of these things.

Josh Robb:
Yeah, I mean, it’s hard. I do think there’s probably room for improvements in financial literacy and education is probably the big thing for me is like, how do we make sure that people are well equipped for this stuff and not too vulnerable to kind of in undue kind of persuasive tactics or whatever else. That’s probably the key thing I’m concerned about is, you know. Yep.

Paul Spain:
And of course we’re in an AI world, so there’s ways in which, you know, folks can lean into AI to do all sorts of things, whether that’s the education or building algorithms and so on to give, you know, to give you something that’s maybe custom to you in terms of helping make decisions. So it’s definitely a fascinating time. So yeah, on the SpaceX front, I have no idea what their future holds, but it looks to be on a little bit of a downward trajectory at the moment from opening on the Nasdaq on, what was it, 13th of June from a New Zealand perspective to—

Josh Robb:
14th.

Paul Spain:
today and they’re around $120 US. So yeah, a bit of a drop.

Josh Robb:
Probably a bit inevitable, to be honest, given how frothy it was at listing. But yeah, yeah.

Paul Spain:
So yeah, interesting things ahead on that front. I think one of the things that maybe doesn’t get Quite enough attention from an investing perspective is how is the world different in 2026 when it comes to the future of businesses compared to when investing? And I even, I remember, you know, I probably had my sort of earliest, you know, dabbles in the share market going back about 20 years ago. And my business partner, you know, put me onto his share, you know, sharebroker, stockbroker. And, uh, you know, I remember calling them up and, oh, I wanna invest in this tech company.

Josh Robb:
Yeah, yeah.

Paul Spain:
And, uh, you know, they had to go away and find out what the actual current share price was because they weren’t even tracking. And I think it was, it was probably like, I remember, you know, there was like AMD, Microsoft, you know, big well-known companies.

Josh Robb:
And they weren’t even tracking.

Paul Spain:
And they weren’t tracking. They didn’t have like the current, the current price. So they’re like, oh yeah, I’ll go and get you a quote for that. I mean, this is so prehistoric. But I remember advice around, oh yeah, you should be investing in these blue-chip firms. And, you know, and some of them, yeah, they’re still, you know, relevant today. But if you, yeah, actually, it’d be interesting to go back and kind of look at all of the, all of the, what, you know, I was being told were the good ones to invest in, like Air New Zealand and The Warehouse.

Josh Robb:
That’d be a good little project for the Discord or something like that.

Paul Spain:
And so on, right?

Josh Robb:
Just do some analysis.

Paul Spain:
What would’ve happened if I’d invested Whereas I think, you know, we have a lot more knowledge today on what sectors and markets have got disrupted, but also there’s still a whole lot of unknown. You know, we look at, say, you know, Xero, our biggest, you know, SaaS company, right? And their stock is at roughly half of what it was a year or 18 months ago.

Josh Robb:
Yeah.

Paul Spain:
a year or so back. And, you know, that all just kind of happened, you know, happened quite, quite quickly. And in fact, probably below, below half of what they were a year. So a year ago they were tracking at around $180 Australian dollars. Today they’re at $68. And this is very common across a lot of those software companies.

Josh Robb:
Those kind of older SaaS businesses or more established SaaS businesses, right?

Paul Spain:
Yeah, it’s, it’s really, really Really interesting. And I don’t know that anyone could say, if we looked at the top software businesses in the world, could give you a great prediction of where these companies are going to be over 5 or 10 years, even whether they’re going to be around.

Josh Robb:
Around.

Paul Spain:
And if they are around, whether they’re going to be stronger or weaker. than what they are today. Like there’s actually that aspect where there’s so much uncertainty of the role of AI and disruption and changing business models. This stuff’s hard, right?

Josh Robb:
It really is, especially in the face of this much disruption. And I think there’s kind of 2 independent things that are kind of converging, right? One is that, you know, you’ve had this crazy weird sort of long period of very low interest rates where kind of return on capital was nearly impossible. And so there’s a lot of SaaS businesses that had quite poor cost discipline, if I’m being honest, both on staff and also on kind of cost to serve, cost of goods sold, whatever you want to think about. And so there’s a bunch of SaaS businesses whose cost structure is probably higher than it needs to be. for the probable underlying value of the product and the service. And so what do you do about that is a really hard problem. That’s real hard not to unpack. And then at the same time, you’re seeing this disruption in terms of, well, actually, you can have a small team do a lot more work than you could 3 years ago, a small capable team.

Josh Robb:
So do we even need all these people? What are they doing? It’s tricky. And then the other question is what is the future of software and SaaS in general? Does all software just become ephemeral eventually? Do you just kind of produce software as you need it and then dispose of it? That’s possible. That’s one possible scenario. In which case, what is the value of a business that sells software? There may still be value there, but it looks quite different. And I think either way, SaaS businesses typically run well over 80% gross margins and it’s inevitable there’s margin compression there. Like, you know, what’s the Jeff Bezos famous quote? You know, your margin is my opportunity.

Paul Spain:
Yeah.

Josh Robb:
And so sooner or later that comes to every sector. And there’s been this kind of golden age of, you know, running 80, 90% gross margins on SaaS. But really, what does a 30% gross margin SaaS business look like? We’re about to find out. And I keep on hearing stories from people about this where like there’s a billing platform that we used at Pushpay called Zuora, which is— billing’s really complicated. And especially in the States where you’ve got like about 1,300 different tax jurisdictions and stuff, it’s very complicated to meet compliance on that. Was public, 4,500 employees. Founder bought it back, took it private. And his goal, his stated goal is to get it to 1 employee.

Paul Spain:
That’s insane.

Josh Robb:
But it’s already down, I think, 2/3.

Paul Spain:
Wow.

Josh Robb:
As he just cuts cost out of the business, cuts out overhead. Because I mean, if you think about SaaS, one way to conceptualise SaaS is that customers effectively pay for acquisition of future customers. And my view as a customer is, well, why should I pay for you to get more business? I should just pay you for what I’m using, not for whatever else you are doing. So I think there’s probably a big structural inversion there, but just how that looks and plays out over the next few years is going to be really interesting. I still think SaaS is a super valuable commodity in terms of I don’t want to have to manage software. I’d much rather pay someone to manage security and compliance and upgrades and security and everything else for me. But I want it at a reasonable price. And if I can build it and host it myself for $100 and you want me to pay $10,000, there’s a big, big, big road for me to hoe to justify myself signing a big check for that.

Josh Robb:
So where is the trade-off in the middle there?

Paul Spain:
Yeah, it is going to be fascinating to see What are the changes? I saw one recently, and I don’t know how long they’ve had this policy and whether it just kind of aligns with your earlier comments, whether it relates to AI, but a company called Halo who deliver information technology service management and professional services automation.

Josh Robb:
Oh yeah.

Paul Spain:
And now CRM, they’re publicly stated that as they grow their customer base and they have more users, that their price will track down. And so they’ve said at this number, we will take 5% off the current price. At this number, we will reduce it another 5%. At this number, we’ll reduce it another 5%.

Josh Robb:
Fascinating.

Paul Spain:
So that to me is setting, you know, quite an interesting standard. The other thing that stood out about their approach in a sector where we often see private equity buy up firms, maybe multiple firms in the same space, Yeah. is they’ve said that they won’t divest, the existing owners won’t divest the business in the next decade. And at this stage, that’s a rolling decade. That’s from today. So in a year’s time, unless they’ve given some other announcement, then it’s still going to be a decade away. Now, there’s probably all sorts of things that could be sneakily done if a business kind of wanted to get out in some ways. I don’t know.

Paul Spain:
But that to me was Those were two things that I hadn’t generally been recognising as a differentiator. And I think they’re really strong.

Josh Robb:
There’s some really interesting models kind of emerging in that space, aren’t there? Around kind of like, okay, well, there’s a number of funds that are, or entities that are buying up software companies and like, we are holding this forever. Like we’re buying and holding it.

Paul Spain:
Yeah.

Josh Robb:
‘Cause customers are sick of, oh, we’ve got a new owner, we’re changing strategy, or we’re de-investing in this, or we’re forcing you off this platform and onto that. And there’s some sort of like, hey, we’d just like some sense of like, we’re the customer here. Yeah, yeah, yeah. So I think some really interesting opportunities alongside some of the challenges. But yeah.

Paul Spain:
Yeah. And it makes the process of selecting software now, you know, more nuanced. There’s more things to think about and to research when you’re helping an organisation, you know, make those sorts of decisions. And that, you know, that’s part of the work that, you know, that I do. And we had one scenario with a US-based, software company. We had a client that was, you know, saying, look, we want to buy this software. Look, just, you know, just get this done and get it implemented, right?

Josh Robb:
Oh, wow.

Paul Spain:
And we were like, oh, well, we, you know, there’s some checks and balances here that, that are quite important from a cyber perspective and a due diligence perspective. And yeah, we managed to convince them, just, you know, slow down a little bit, let us, let us do some of that work, work in the background. And so So we fired up our process and I was sitting on the call to the salesperson in the US and we got about halfway through the call and he’d been just touting why we definitely should buy this bit of software. And I said, okay, now in this sort of day and age where private equity firms buy up multiple competing products and so on, what’s your story?

Josh Robb:
Yeah.

Paul Spain:
I didn’t know, I didn’t know at all at that point. And, you know, that was part of the reason for being on the call is to fill in the picture. And he said, oh, I was going to tell you, Paul. Yes, we’ve got this one product that we’ve stopped development on, the product you’re looking at. Yeah, we’ll be finishing up development on that soon. And then we’ve got this other one. That’s our cash cow. That’s what all the focus is going to be on, right? And if I hadn’t made that call and we’d just said to the customer, look, if you’re convinced, if that’s what all your other, you know, related firms are saying is the best product to use and so on, okay, well, you’ve done the research, let’s just do it.

Paul Spain:
Yeah. I mean, it’s, you know, a huge decision, key, you know, key part of their whole operation. They would have gone down a really bad, really bad track.

Josh Robb:
So the other thing I’m seeing at the moment is kind of incumbent legacy products that have that kind of captive market. It sounds like that’s the scenario you’re talking about a little bit, specialist kind of products in a vertical who are also completely interoperable and their strategy is to not allow any other players in. So there’s no APIs, there’s no integration. And I think in this current world where it’s becoming very quickly possible to add capabilities to software using agents or using AI or whatever, Yeah. that actually the kind of openness of the platform becomes a really important decision-making tool because you can end up painting yourself into a corner really quickly if you don’t get the right rights to access platforms or data or whatever as part of that procurement decision.

Paul Spain:
Yeah, that’s really challenging. In fact, we had, you know, JD Drask on the show recently and, you know, we were talking about a challenge with Xero. Yes, they’ve, you know, they’ve been amazing in terms of having APIs since year dot and really kind of leading the way on in building an ecosystem, but now they’ve changed how they build for that. And if you want to use your GenAI platform to talk to Xero, is there an MCP capability? Well, yeah, kind of. They’ve made something available, sort of open source that you can download. It’s not a native functionality where you can just chat to it from Claude or what have you.

Josh Robb:
Yeah.

Paul Spain:
And yeah, he was certainly sort of bemoaning some of the challenges there. But look, I get it. You know, when you look at established businesses—

Josh Robb:
Incumbency, right?

Paul Spain:
That ultimately are incumbent, right? They are now the incumbent, not the disruptor.

Josh Robb:
Yeah, exactly.

Paul Spain:
And so we’ve got the disruptors trying to come through. they will be trying to hold on to their place and their profits and so on. And but yeah, in some ways it reminds me of kind of looking in at Sky TV in New Zealand a few years ago.

Josh Robb:
Yeah.

Paul Spain:
And occasionally I got, you know, quoted in media for throwing stones at them. But, you know, look, I could also understand they had a, you know, a really, really strong revenue base, which, you know, to a degree has has continued. However, yeah, if you don’t adjust, then yeah, you will get, you know, these sort of incumbents will get knocked out of the market. And look, you know, Sky have kind of somewhat reinvented themselves and they’re still around and still relevant. You know, maybe their valuations are not—

Josh Robb:
Quite what it once was.

Paul Spain:
what it once was. So, you know, I sort of, I think it is important to sort of step back and to look at these other perspectives as well. Yes, there’s always going to be a customer perspective. And as a customer, you can always choose to—

Josh Robb:
Change.

Paul Spain:
Do something different. But, you know, there’s a responsibility to the shareholders as well that, you know, there is thought that goes into things, not just, hey, we’re moving with this trend because, everyone else is doing it. And so I know these things aren’t easy. I know we have a lot of people in New Zealand who, you know, are part of Xero and we want all of them to keep their jobs and Xero to employ more people.

Josh Robb:
Right.

Paul Spain:
How that plays out, time will tell.

Josh Robb:
Some hard problems. Yeah.

Paul Spain:
A couple of other bits in the news. AWS had a bit of an issue. when it came to their, I guess, their engine that tracks what your spend is. And we saw folks getting crazy, crazy bills. I think in one case somebody got a bill for, or an estimated invoice total for the month coming in at $1.5 trillion.

Josh Robb:
I saw a billing consultant say they’d seen one customer with $23 quadrillion Oh no, that’s crazy.

Paul Spain:
So absolute mayhem. It did leave me curious, and I don’t know what you read on the topic, but as to— has someone sort of vibe-coded, uh, something in there to do with their estimation engine without having that thoroughly checked, or is that a completely human mistake?

Josh Robb:
But, uh, I mean, Amazon’s an amazing— or AWS is an incredible organisation. And I, you know, there’s 2 services that AWS run that just absolutely blow my mind. And one is the Amazon Billing Service.

Paul Spain:
Yes.

Josh Robb:
And the other one is the Amazon Identity and Access Management Service. Those 2 services are at the bottom of the dependency tree. Every other service they’ve ever built depends on those 2 things, and they have to be 100% accurate and 100% available all the time. And so it’s just fascinating to me that they even scale when you consider like they’re metering milliseconds of Lambda usage and everything else.

Paul Spain:
Yeah.

Josh Robb:
But I mean, this is obviously pretty public. Who knows? The quality has been a bit more variable recently than it had historically been. And they’ve seen a lot of kind of senior departures over the last kind of 3 or 4 years. So it’s concerning.

Paul Spain:
Now, 2 other topics I wanted to tap on quickly before we sort of delve in for an update on TENT. One was the crazy situation that we’ve kind of landed in when it comes to, you know, to zero-day vulnerabilities or cybersecurity issues that are happening and being taken advantage of in the real world before, you know, fixes are available publicly. So Microsoft have been caught in that world and so has WordPress. Microsoft has this month addressed more vulnerabilities in a single month than they ever have done in history before, something like 600 vulnerabilities addressed in the last month. This compares with normally they’re patching maybe 50 to 100 vulnerabilities, so 5 to 10 times what is your typical month. And so that creates a pressure for organisations to get those patches out, get them deployed very, very quickly.

Josh Robb:
Yeah.

Paul Spain:
very quickly when some of these things are actually actively being exploited in the wild. And then there was WordPress Saturday New Zealand time. Yeah, we had our team rapidly getting patches out for, you know, organisations that we work with that were running WordPress because there was a pretty shocking Vulnerability there. You could go to, I think, the simple breakdown, you could go to the login screen, inject certain content into the login that would basically allow you to take over and take complete control.

Josh Robb:
It’s a bad day for WordPress.

Paul Spain:
Of a website and the web server. Interestingly, WordPress, this is an open-source platform. they initiated a forced patching of customer web servers, which is virtually unprecedented. It is very, very unusual to see this type of action where, you know, somebody who you don’t have any sort of, you know, commercial partnership with or whatever.

Josh Robb:
They’re not hosting your stuff. Yeah.

Paul Spain:
And they’re you know, I guess the foundation behind WordPress. I haven’t looked up the legal bits and pieces there, but they made the decision that with 500 million installs of WordPress running around the world, if they didn’t actually trigger those updates, then there might be even worse news.

Josh Robb:
More problems.

Paul Spain:
But I’ve seen very little coverage of the fact that they actually took control of people’s web servers and Yes, quite likely knocked off some, you know, broke some things and so on, but much less damage than there would have been otherwise.

Josh Robb:
Yeah. I mean, I think it seems like the brand damage would be— it’s funny, I actually had missed that aspect of it, which is interesting. And it’s interesting they’ve even got the capability to do that.

Paul Spain:
You know, I didn’t know they had the capability.

Josh Robb:
Yeah. Yeah. I can imagine how it might work. Yeah, that’s fascinating. I mean, I think it’s actually probably sensible if the alternative is, you know, $100 million you know, compromised WordPress installs. Like, that’s a bad day for a lot of people, but—

Paul Spain:
Depends on their governance. Yeah, because you could imagine if a bad actor had taken control of that, shut it down, and, and did, you know, decided to take control of 500 million websites, then we would have had some different news headlines.

Josh Robb:
Yeah, what’s the vulnerability in that one, right?

Paul Spain:
So yeah, fascinating times. I think a key reminder there is as a country, we’ve just got to keep stepping up from a cyber perspective. Anyone that’s involved in ensuring their organisations are patched and kept up to date, our job from that perspective is even harder today than it was yesterday because we need to keep these things moving really fast in an AI—

Josh Robb:
Yeah.

Paul Spain:
world where—

Josh Robb:
Although there are benefits to that too, right? And I think it’s a double-edged sword, that one, because on the one side, I mean, part of this is being caused by AI. You know, what’s Linus’s thing? You know, in the face of a million eyes, all bugs are shallow. And so if you can use AI to discover vulnerabilities and you really rapidly increase the rate at which you find them. And so the question is, are the defenders or the offenders—

Paul Spain:
Winning.

Josh Robb:
Harnessing that first? And at the moment, the state-of-the-art models have these guardrails, which mean that you can’t use them for cyber defense. So, but, you know, just over the weekend, KIMI-3 came out, or K3, and that model is probably as, like, it’s somewhere, it’s very close to, if not at the state of the art between Opus 4.8 and Fable. And there are no guardrails and it’s open weight, so you can run it anywhere you’ve got inference compute.

Paul Spain:
Yes.

Josh Robb:
So all of a sudden people now, like both offenders and defenders, have a capability that they can use to scan and improve stuff. I was just— I spent an hour talking about this with my team this morning. We’ve just done a pen test on a new series of products that we’ve built. Got back a completely clean pen test from the pen testers, and pen testers hate having zero findings. Like, there’s a really crappy— Feel like a failure. Yeah, they really hate it. They hate it. But the team used AI to do all of the prep for this.

Josh Robb:
So while they were developing this new set of products, they were scanning— the AIs were scanning them, giving them feedback. making recommendations, fixing some of the stuff themselves. So they had quite a high degree of confidence going in that we were in a good place. And the pen testers are just kind of verifying that, which is like, so, you know, like the defenders also have these capabilities, which I think is a really encouraging thing. So I think—

Paul Spain:
Yeah.

Josh Robb:
It’s going to be a real rocky year, but I’m actually quite optimistic about the overall improvement in software AppSec as this evolves. What’s going to be hard is the organisational institutional responses to these, because that moves much slower than the technology. Yeah. And particularly if you’re like a, hey, we patch once a month kind of an org.

Paul Spain:
Those days are over.

Josh Robb:
Those days are well and truly over.

Paul Spain:
You have to be able to move fast and get things out. And we will probably see more of the, you know, sort of out-of-cycle, you know, patches coming through. And yeah, we just have to learn to roll with that. And look, yeah, there’s some risk because there’s not enough time to actually test everything that you kind of have to roll it out at times.

Josh Robb:
And there are some really interesting products emerging around this as well. There’s a New Zealand startup called RedShield. Have you seen them? And they they have a WAF that sits between your applications and the internet and effectively adaptively responds to incoming threats and blocks them. So if you can’t patch the underlying system quickly enough, you can patch it at the WAF level, which is, I mean, that’s instant. So that’s like, I think there’s a lot of really interesting adaptive, and I know they’re doing some stuff with AI at the moment around kind of real-time response stuff. So I think there’ll be a whole nother set of capabilities that defenders get that will be quite powerful as well. But it is a very uncertain time.

Paul Spain:
Yeah, yeah.

Josh Robb:
And there’s a need. I mean, I think you said this, but, you know, there is a real need for governments and institutions to make sure they’ve got good programs here, in particular New Zealand. Like, certainly what I see is a lot of underinvestment in this space.

Paul Spain:
Oh yeah, this is what we were chatting about before this podcast.

Josh Robb:
Exactly.

Paul Spain:
Yeah, yeah, yeah, yeah. There is that. There’s that. Yeah, that difference that probably New Zealand has had compared to, you know, say the NCSE and the government guidance in the UK for differing types of organisations and how they should be protected. And yeah, the more that as a country we can take some learnings.

Josh Robb:
Yeah. Well, and even just kind of, I mean, I think there’s going to have to be, you have obligations around health and safety and you have obligations around other things. Like you should have obligations around your cyber hygiene or your cybersecurity to some degree without maybe too much regulation, but there needs to be some sticks otherwise.

Paul Spain:
Yeah, well, I think it’s getting easier too. Now, just one more topic quickly. Japan has decided to make what is being referred to as a bold national step on AI, partnering with the likes of NVIDIA to build a sovereign AI infrastructure capability focused on physical AI technologies to power robotics, industrial automation, digital twins. Yes, they’ll be able to do, you know, I guess more standard GenAI type things as well. And they’re saying this project will provide computing, a computing foundation for Japan, the Japanese government backed Front IA Frontier initiative, bringing together manufacturing expertise, industrial data, and advanced AI infrastructure to accelerate innovation across a range of sectors where Japan has a level of strength. And, you know, I think that they see this as an opportunity to I guess, you know, future-proof themselves a little bit by making that investment now. So they’re talking about an AI factory with 140 megawatts of dedicated AI data center capacity. And was it 27,500 NVIDIA Rubin GPUs? There’s a whole lot of money there.

Paul Spain:
That’s a lot of money. Being spent into the, you know, their own capabilities. This is really interesting because, you know, I guess most countries have been, you know, just, look, we’ll leverage what the US is doing.

Josh Robb:
Yeah.

Paul Spain:
And, you know, it’s kind of been US and China where most of the investment’s gone. There has been a chunk of talk around what could and should sort of sovereign AI investment look like? And well, here’s one approach.

Josh Robb:
I mean, it’s interesting because, I mean, obviously Japan has this big demographic cliff, right? So they’ve got a massive workforce shortage baked into their demographics. And so they’ve been big in robotics for a while because they’ve been looking at how are we going to care for old people if we don’t have enough young people to staff facilities or hospitals or aged care facilities, whatever it is. So it makes total sense to me that they’re focusing on that kind of robotic approach. They’ve already got deep chops there and they know they have to solve a real economic problem for themselves in terms of how they maintain their manufacturing base and their service sector. So that makes real logical— that seems like a very logical strategy to me as opposed to, oh, we’re just going to like put up a data centre and hope that it turns into some economic outcome.

Paul Spain:
Yeah.

Josh Robb:
You know, we think we can make better capital allocation decisions than Silicon Valley seems like a poor bet for a state. without some grander strategy like that. So, yeah.

Paul Spain:
Yeah. And it’ll be interesting to see how others look at this and, you know, what approaches other countries with deep pockets decide to do. Now on to Tend, keen to hear where things are at. It’s been a while since we’ve had you on the show. Lots of things going on. Now I understand that Greencross Medical is, you’re close to sort of finalising that acquisition, but maybe for those who aren’t familiar, you know, what is Tend? What’s your capability and your spread around the country and so on?

Josh Robb:
Yeah, great. So we’re a primary care provider or a healthcare service provider. So we have about 500 staff. Mostly doctors and nurses and healthcare assistants and receptionists and that kind of stuff. So very much a services business, but really digitally enabled. And it’s been built out as a platform. So all of our team are on a single instance of everything. We can move work around, we can move phone calls around, just a single instance of people’s patient files.

Josh Robb:
This sounds really normal to people that are used to the knowledge economy and knowledge work, but within healthcare, everything is siloed.

Paul Spain:
Yeah.

Josh Robb:
Everything is siloed and it’s been really, really hard to kind of work through breaking down the walls. But the idea is that we want to make it really, really easy to get access to the kind of care you need when you need it for a consumer or a patient. And we want to try and make the job of providing healthcare services for our workforce easier as well, because that’s really hard. And when you look at— I mean, you would see this if you go and visit your doctor, but you look at the systems they’re using and you look at what they’re having to do, and there’s a lot of clicking and a lot of typing and a lot of things that look like they haven’t been touched in 20 years from a technology And you wonder how secure they are as well as how productive they are, right? Exactly. So we’ve kind of, we’ve built a platform for the business that’s a set of products with partners that we have really great relationships with, some technology that we’ve built ourselves, and we’re working really hard as fast as we can to integrate that so it’s easy to use for the workforce and the information’s joined up. And there’s always been a strategy since we started Tend to use AI. What’s really cool is the capabilities have come online with really good timing. But we actually did a Callaghan Growth Grant, or project grant, sorry, in 2021 that was all about like, okay, the first thing before we can do anything clever, we have to instrument the patient interactions and the provider interactions to just start actually understanding what’s happening so that we can then work out, okay, how do we start moving work around? Where can we automate? What are the opportunities? So we’ve spent the last kind of 5 or 6 years doing that.

Josh Robb:
And the way that we’ve done that is by buying medical centres and then doing quite intensive kind of transformation on them. So full change management in terms of people’s role. Well, I mean, the job of a doctor or nurse is the same as it always is really, but really changing some workflows, changing the operating model quite dramatically, changing models of care, introducing new technology, joining it up. And that’s really difficult because often these people have been working there for decades in some cases, and no one loves change. Even the people that say they like change don’t really like it when it happens to them. Sure. But buying medical centre, so we’ve got 25 medical centres at the moment currently, and we’ve acquired some of those as small networks in groups, and some of them we’ve acquired one at a time. We’ve integrated 15 of them into the 10 networks so far, so they’re all joined up.

Josh Robb:
And then we also offer these online services, like we have a telehealth service, we’ve got a service that lets you order medical certificates if you need one, and a bunch of other stuff.

Paul Spain:
Making it easy for people to have a sickie?

Josh Robb:
Yeah, well, I mean, it’s not so much making it easy, it’s more, I mean, like—

Paul Spain:
I’m just having a chat.

Josh Robb:
I’ve never been asked for a medical certificate in my career.

Paul Spain:
Yeah, likewise.

Josh Robb:
But if you work in a factory, you’re almost certainly gonna be asked by your employer for a medical certificate. And so it’s really unequal how that works. I remember talking to the chair of an iwi, Ngāti Maniapoto, and she was telling me that their iwi members, if they get sick, they work at a factory, their only option is to go and stand outside the medical center at 7 o’clock in the morning while they’re sick so they can get a medical center so they don’t have to go to work, get a medical certificate so they don’t have to go to work. inequitable and unnecessary. And the other thing is that doctors are not like— it’s a really bad misallocation of a doctor’s time to do a medical certificate. That’s all you need. It turns out nurses can sign medical certificates from a legal perspective. So we use nurses to do that, and that works really, really well.

Josh Robb:
There’s really good scaffolding. If they’re concerned about your wellbeing, they can refer you straight into a doctor, which is really cool, telehealth. But just makes things move more easily. We can do that at a much cheaper price. We can do it more conveniently, et cetera, et cetera. So At the moment, you can see a doctor on our platform via telehealth normally within 2 or 3 hours. And we work really hard to get in-person appointment availability down. It’s complicated because humans are humans and people get sick and whatever else.

Josh Robb:
But normally you can get an in-person appointment in a couple of days across our network, which is really cool. But growing the network by buying medical centres is slow and painful. And those transactions kind of, they just, they eat a lot of time and capability. We’ve entered into an agreement with Greencross to buy the Doctors Network. The shareholders, the Greencross shareholders approved that last week. And we’ve done that with support of a whole bunch of iwi investors, which is really exciting. So we’ve had really great support from iwi over the last few years, but we’ve got a whole bunch of them as capital partners for this transaction joining our shareholder register, which is really exciting.

Paul Spain:
That is. And it’s a little bit extra exciting for me because I’m— I’m on their patient—

Josh Robb:
You’re a doctor’s patient.

Paul Spain:
On their patient register. And look, yeah, quite interesting from the technology perspective. It’s always felt like they were backwards. And even, yeah, I remember having a discussion with one of their GPs about something technology-wise that wasn’t a good indicator. And it was like, wow, we’re now part of, this group and we’re owned by Greencross and that means, you know, XYZ in terms of getting, uh, you know, getting, getting changed. So yeah, it will be interesting to see, you know, how, how these things all kind of, uh, you know, move and, and progress over time. , interestingly, on the, on the, on the, on the— and, oh, get— yeah, getting, um, uh, repeat prescriptions. I remember that, that being, um, that being a bit of a challenge even though Yeah, they had, you know, Manage My Health, et cetera, et cetera.

Paul Spain:
I think they moved on from that, but they didn’t remove me off Manage My Health, which I had some struggle to get any—

Josh Robb:
Struggles, yeah.

Paul Spain:
Any sort of accountability, you know, from the local outfit on, well, you put me here, don’t you take me off? And sort of an unwillingness to respond. But on the flip side, They seem to have this thing where you can turn up on a weekend and usually always get an appointment.

Josh Robb:
Yeah, nice.

Paul Spain:
So just kind of a walk-in type scenario. So don’t flush that down the toilet. Don’t turn it off.

Josh Robb:
You’ll have one unhappy customer. Yeah, no, that’s great.

Paul Spain:
But anyway, no, that’s really good. Now I’ve noticed Yeah, an increasing number of, of TEND, you know, related clinics around. Yeah, we’re working hard.

Josh Robb:
Not just Auckland.

Paul Spain:
So that must be quite a big job to take an existing, you know, clinic that gets acquired, or group of clinics, and then need to, you know, kind of pull them into your model in terms of data or in terms of all those changes that you were you were talking about in terms of how the operations actually work. How hard is that to do?

Josh Robb:
Oh, it’s really hard. I mean, it requires serious intestinal fortitude, to be honest. People don’t like change. We’ve got a really, really good, very well-structured process for this. It’s kind of a, probably a 14-week program that we take practices through and we run workshops with them. We’re like, hey, here’s how your cheese is moving. This is how your model of care is changing. Here’s the new technology we’re introducing.

Josh Robb:
This is what it means This is how we’re gonna train you on it. And we’ve gotten better and better at that as we did it. So the first, in 2020, when we first launched, we opened a medical centre in Kingsland that we built from the ground up. And then we acquired Simon Street Medical Centre the next year, end of the next year, and kind of did that as well. That was a bit of a turnaround job that had kind of, was going out of business and so it needed some help. But then we started buying big medical centres and big groups. And, you know, so we did Pakuranga in ’23, And then we did 4 medical centres in Tauranga at the beginning of ’24. And then we did a whole bunch in Canterbury, ’24, ’25, to get up to 15.

Josh Robb:
But it is hand-to-hand. You know, we put a team on the ground for kind of 3 to 4 weeks around that actual major transition. And they’ll be back there beforehand. They’ll be there, you know, probably for a week at a time for kind of 3 or 4 weeks in the lead-up to that. So it was very, very hands-on. Yeah. We have a really good continuous learning culture, so we’re really good at kind of reflecting, going, okay, what do we learn? What do we do differently? How our partners perform? Do we need to change any processes? You know, whatever else. But it’s difficult.

Josh Robb:
And I think we’ve got better and better at making the surface area of the change smaller for various groups. So doctors these days, for the most part, are pretty— like, it’s probably the smallest surface area of change for them. They change the system that they’re using, but broadly we’ve done a really good job of making that very, very simple for them. We introduce some new things, but those new things are generally like, we introduce our in-house AI scribe, which actually makes their life easier. So there’s some kind of, there’s some treats for them along the way. For nursing, the change is a bit more, but again, I think we’ve got, and we always use subject matter experts. So the people that we bring in to help do the change are doctors that are working in centres that have been through this process. The nurses that we bring in and nurses that have been through this process and centres that have been through it.

Josh Robb:
So it’s kind of like, hey, I’ve had this happen to me, this is my experience, this is what’s going to happen to you. The first 3 weeks are going to suck, you’re going to hate it. , sometimes people cry, occasionally people, you know, get very frustrated, you know. But, and so we get good at de-escalating people and whatever else. Sometimes I think, man, I used to write software for a living, that was way easier, you know.

Paul Spain:
, dealing with computers versus people.

Josh Robb:
Yeah, but, you know, I think like it’s It’s been really powerful. And as I said, you know, it gives us these really powerful capabilities in terms of moving work around, taking toil. There’s a lot of toil, particularly that GPs and nurses do, that is just busy work that can be automated. So we have some RPA that we run that does a lot of our administration that’s just filling in forms.

Paul Spain:
Is it robotic process automation?

Josh Robb:
Yeah, sorry, robotic process automation. Yeah, so it’s just literally filling in the forms that you have to fill in. So that like takes that entire job. There’s normally someone’s job in a medical center whose job is to kind of file the paperwork with ACC for claims, and that’s like a whole job. And they normally come in in the morning, there’s a big stack of paper from yesterday, and they kind of work their way through and make sure they’re all coded correctly and they’ve been submitted. That turns into a button.

Paul Spain:
Fantastic.

Josh Robb:
You know, so, so there are upsides. The challenge probably is that the health information systems have been really sorely neglected over the last 20 years. New Zealand used to lead the world in some of this stuff. We were the first country anywhere that had like a national health index unique identifier for health. , we have some other cool capabilities, like our e-prescribing system is, is pretty good. But overall, the governance and the curation and the roadmap around that has, has not been well managed by, by the health system custodians. And we’re in a bit of a difficult situation now in terms of how fragmented it is. , so it’s, it’s not straightforward to just kind of wave technology at it because a lot of these things don’t have APIs.

Josh Robb:
A lot of them are like running horrendously ancient technologies, um, and, and there’s a lot of investment required, but we’re making really, really good progress on it. And I think there’s clearly a path to a system that is much easier for the workforce.

Paul Spain:
Yes.

Josh Robb:
And if you can’t make it easy for people to do the job, then you can’t deliver the services and the care that you need to the community. So it’s really about that virtuous cycle. Even, I mean, we have a, I think, 80-something percent of all of our eligible patient interactions use our AI Scribe. So that’s something you can send into as a patient, and then the doctor will use it and it will produce clinical notes. It produces a really nice patient instruction for you written in the doctor’s voice at a kind of 14-year-old reading age, no jargon. It says, hey Paul, this is what happened today, this is what we’re doing, here’s the plan, here’s any safety netting or red flags you need to look out for, this is what you need to do next. No doctor in the world has time to write that. And one of our really experienced GP fellows said to me the other week, he was quite emotional, he was like, man, Josh, this just makes me such a better doctor.

Paul Spain:
Yeah.

Josh Robb:
Like, in my heart, I would love to do this for every patient, but I don’t have the typing skills or the time to be able to do it. But the fact this is done for me automatically as an outcome of an encounter using AI is just incredible. And that’s really powerful. And the other thing that that gives us is it gives us this— every time you see us, the quality of your clinical record improves, which is not something that normally happens in primary care. Because normally it’s a busy doctor.

Paul Spain:
Right.

Josh Robb:
I write a few lines in kind of cryptic hieroglyphic And it’s really just to remind me what we talked about. It’s not useful to you as a patient at all. It’s not useful to nursing or anyone else that needs to interact with you. And so if you then call up and say, hey, it’s Paul, I’m just following up, like, I got given this medication and da da da da, you’ll probably talk to a nurse because the doctor will be busy. I mean, the nurse has to go and knock on a busy doctor’s door, interrupt them, maybe get their head bitten off if they’re having a difficult day, and then be like, hey, Paul needs to know. Whereas now they’ve got this really great clinical summary of the encounter.

Paul Spain:
Yeah.

Josh Robb:
They can see the medical reasoning and they can be like, oh yeah, cool, like I can see what happened. Yeah, cool, like do this or don’t do that. Or okay, actually probably need to come back in again, but they can make a really safe decision. And so the feedback from our nursing teams is, hey, this has really helped us provide better quality care to our patients as part of a true multidisciplinary team as opposed to the GP being the kind of singular locus of primary care.

Paul Spain:
2 things come to mind. for me. One is, it’s been talked about for a long time, but having a medical healthcare system that is actually proactive.

Josh Robb:
Yeah. Yep.

Paul Spain:
Do you see the way that things are coming together with the way that Tend operates, that actually you’ve then really laid the foundations for a completely proactive healthcare type capability because you’ve actually got data in a kind of useful way. There’s other bits and pieces I’m sure that have to kind of be added in to be truly proactive, but it sounds to me like it’s kind of some of the foundations that will allow that integration. Yeah, I mean, I think we’re on the right track, right?

Josh Robb:
So I mean, I think that, what’s the old adage, garbage in, garbage out. The quality of data in healthcare in general is pretty poor for a whole variety of reasons. Like if you trained as a nurse or a doctor, like you’re not, you haven’t trained to be a data entry clerk, right? And it’s not the highest priority for you to make sure that I’ve coded all of these things about this encounter correctly.

Paul Spain:
Yeah.

Josh Robb:
And so, but AI is really good at pulling structured information out of unstructured information. and making categorisation decisions. So we’ve got a bunch of agents that are working to kind of like go, okay, like let’s just look at the transcript of this encounter that Paul’s had with the doctor, and let’s just look at his problem list and the long-term conditions he’s been diagnosed with and go, which ones are relevant to today’s encounter? And that’s really helpful for us because it lets us do what’s called case mix. So like, why are people visiting? And once you can do that, then you can say, oh, actually what we should do is we should start a diabetes clinic. or a weight loss clinic, or actually we need to get some more mental health resources into this location, because actually a lot of what people are seeing GPs about is mental health. GPs can’t do much for mental health in 15 minutes, but if we put some health improvement practitioners in there, counselors, they could be spending 30 minutes with Paul helping him with his anxiety issues or his relationship problems or whatever it is. And that’s a much better use of everyone’s time.

Paul Spain:
It’s a— We’re not supposed to mention those publicly on air, Josh.

Josh Robb:
You know, But it’s kind of that kind of thing. So it’s like, how do we get the right— how do we get you as a health consumer in front of the right person for your needs today?

Paul Spain:
Yeah.

Josh Robb:
It’s a really hard problem. And in a really well-run medical centre, the receptionists are the unsung heroes of this. They help you navigate. They’re like, oh, look, Dr. Bob isn’t available today, but we can get you in with this person and they’ll help you. Or actually, look, don’t worry, it’s going to be okay. I’ll put you through to a nurse and the nurse can do some reassurance and say, actually, if you just sleep on this, it’s probably going to be okay and we’ll see you tomorrow. Or, hey, you know, bring your sick kit in and we’ll squeeze you in.

Josh Robb:
Or there’s a weekend walk-in clinic or whatever.

Paul Spain:
Yeah.

Josh Robb:
But how do people navigate? And so getting— but getting the data right and getting the data capture right is really, really critical to that. , so I think we see heaps of opportunity there, but it starts with getting that foundational data right. We need to know about you, we need to know what you’ve got, what you need. I mean, simple things that we do like When we’re prescribing, when doctors are prescribing for you, or any prescriber is prescribing for you, they’re prescribing you certain drugs at certain quantities and a certain dosing amount. And so like, we know, like, if we’ve prescribed you 2 pills a day for 30 days, we know if you’re taking it, if you’ve got perfect adherence, if you’re taking it twice a day, we know how many days supply you’ve got. You should never run out.

Paul Spain:
Hmm.

Josh Robb:
So we send you a push notification about 5 days before you should be running out and say, hey Paul, are you running low? Do you need to get a repeat prescription? Click here and order more. Like, just basic stuff, but you have to have all the underlying data quality right to be able to do that in the first place. And that’s been a really hard lift, but we kind of— we’ve broken the back of a whole bunch of those things. So yeah.

Paul Spain:
Yeah, yeah. Uh, there’s, there’s so much more we can kind of go on, on this, uh, topic, so we’ll, we’ll have to— but we’ll have to save some of that for next time. , in New Zealand especially, after you know, what happened with ManageMyHealth, there’s naturally going to be an interest in, you know, how do you ensure that, you know, data is kept secure? How what somebody expects to be private is kept private? How are you kind of, you know, tackling that on a day-to-day basis? Yeah, that’s a really good question. And I like what you said earlier around, you know, the new code that you were building with AI. And that you’re able to build in some really useful checks and balances through the software development process that’s meant when testing afterwards, actually it’s come up very, very strong. But yeah, how do you look at that? Because I mean, it’s a big picture, right? There’s a lot to it.

Josh Robb:
I mean, so my last business was Pushpay and that’s payments, highly regulated. The PCI DSS is this big behemoth of a compliance program.

Paul Spain:
Yeah.

Josh Robb:
, and so I was really shocked when I got into health and there was almost nothing on the technology side. Like really, you know, there is almost nothing in terms of compliance obligations around health technology in New Zealand. How, you know, in other jurisdictions you have regulation around software as a medical device, and in the US they have HIPAA, the GDPR covers some stuff in, in Europe around this as well. New Zealand has the Privacy Act, but that’s a very general piece of legislation.

Paul Spain:
Yeah.

Josh Robb:
And there’s nothing else. And I think Manage My Health and then, and then the other breaches that happened this year have kind of forced a really big systemic response. So there is, um, actually as of July, there’s a new contract in place for primary care. The details are just being finalized that will roll out, um, over the next year that will require health service providers to meet the NCSC kind of minimum requirements, which I think is a really, really, really good thing. And we already do all of that. Like, we run an internal program that looks very much like you’d see in, in a software business or whatever else.

Paul Spain:
Yeah, yeah.

Josh Robb:
But, but to require, you know, individual medical centers to make sure they’ve got the right controls in place or the right agreements with their suppliers for backups and password policies and multi-factor authentication, that kind of stuff is really important. And I mean, historically, there is a tension in healthcare between security and service delivery. You know, it’s not desirable for someone to die because someone’s having to enter an MFA code in an emergency department. So you do kind of have to make some trade-offs there. But I think the bar had been way too low for too long. Earlier this year, we rolled out mandatory MFA for all of our staff everywhere. So up until, I forget when this was, May, you needed to use MFA if you were offsite. But if you’re in one of our locations, you weren’t challenged for MFA, which was a kind of pragmatic decision.

Paul Spain:
Yeah.

Josh Robb:
But we looked at that and we were like, look, we are not comfortable with this. And so we worked with our suppliers and our vendor. And so we rolled out a model where you’ll still get an MFA challenge once a day, even when you’re in one of our locations. And we, ’cause we just think like actually protecting people’s health information is part of our duty of care. It’s not separate to making sure that we don’t hurt you or that we, you know, provide you the medication you need or, you know, access to services or whatever else. It’s part of the same thing. So, you know, Health New Zealand and the Ministry are doing a lot of work on this at the moment, and it’s really encouraging to see them taking kind of bold steps forward. It’s gonna be a big, big lift for the sector, and there’s a real lack of capability within the sector around some of this stuff.

Paul Spain:
Mm-hmm.

Josh Robb:
You know, health doesn’t attract great technologists, frankly. And the people that work there work there ’cause they’re really interested in health, which is awesome, cybersecurity specialists most of the time. I mean, there are some great people out there, but they’re few and far between. And so there is a real need to be able to bring some capability into the sector. So yeah.

Paul Spain:
And then lastly, you’ve touched a little bit on some of the AI usage that you’ve been— the ways you’ve been leveraging AI within Tend. Can you maybe just kind of walk us through that? Because Yeah, this stuff is moving so fast. So a conversation that we had, you know, 3 months ago is like a 3-decade-old, you know, conversation. So we have to kind of keep tapping in on what folks are doing.

Josh Robb:
So I think there’s kind of a few big areas where we’re really leaning in. So one is health information is lots of unstructured data fundamentally. You know, I write notes about— if I’m a doctor, I’m writing notes about your visit. I’m saying, you know, this is what we talked about and this is what we did. And then there’ll be some structured data around what I prescribed. , but that unstructured data is actually really, really rich source of signal if you can use it. So we’re doing 2 things. One is we’re using AI to help draft those notes for doctors.

Josh Robb:
So we’re transcribing conversations and then drafting notes and patient instructions and other things. But we’re also using it for other other sources. So like one example is medical centers get a huge amount of phone calls. And so we want to understand like, A, how well are we serving our customers on the phone? You get a lot of complaints about phone calls. We had to wait too long. I didn’t get what I need, whatever else. So we’re taking a sample of our phone calls, which are recorded in our phone system, running them through a model, and we’re scoring them on a whole bunch of domains like, you know, what was the sentiment at the start of the call? And then what was the sentiment at the end of the call? What was the call about? What do we think the call was about? , did we identify the patient safely? We have like a— we require our staff to do 3 points of identity verification to make sure you are who you say you are. So it’s normally name, date of birth, and address, or phone number, or something like that.

Josh Robb:
Are they following that protocol? Are they following our brand guidelines about how we want them to introduce themselves and describe the service that you’re called up about? So, and then we’re aggregating all of that information so we can use it for, for coaching and QA. It’s funny because 10 years ago at Pushpay, we had a team of people with headphones that would listen to call recordings and then write scoring into Salesforce kind of scorecards for sales managers to kind of go and give feedback to their teams. But now it just costs 3 cents to do 1,000 of them sort of situation, which is incredible. And then the other thing that we’re doing, which is really cool, is we get a lot of feedback from patients. We ask for feedback after appointments and people write comments. We get Google reviews. The Health Quality Safety Commission do a quarterly survey of people that have visited the GP and ask them for feedback. I’m probably forgetting some sources.

Josh Robb:
There’s lots of different sources of kind of unstructured kind of verbatim data. What we’re doing is we’re feeding all of that through models and using it to summarise what we call the voice of the customer. So we do a quarterly report that says, hey, this is what we’re hearing. And some of that’s about service issues, some of that’s people being happy about things, some of that’s missing gaps in our product. Hey, look, you know, we want this feature, we want a web-based version of the Tend app, which is the thing we’ve been working on for the last 3 or 4 months. Or, hey, like, I really want to be able to book like a mental health appointment online and we can’t do that at the moment. Or, hey, I’m frustrated that I have vision issues and the app’s not available in dark mode. So we’re using that to kind of summarize out what we’re hearing from our customer and use that with our executive team and our operational teams to think about how we’re planning services, what we’re focusing on from a strategy perspective, and just really being in intentional about how we respond to feedback from our users.

Josh Robb:
So yeah, that’s been super powerful. , that’s great.

Paul Spain:
Yeah, you, you, um, you mentioned your sort of Scribe, um, you know, transcription AI kind of capability, um, with your, your clinicians, your GPs. Uh, you know, there was a window of time where, yeah, no one was kind of officially doing that type of thing in New Zealand. The technology obviously has, has continued to move at quite a pace. There’s some, you know, quite specific products in the market, but, you know, generally they’re, yeah, they’re largely, you know, using, you know, off-the-shelf technologies with, you know, I guess some, some, um, you know, customisation to, to make that work. What, what are you seeing in terms of how good the, um, the AI transcription capabilities are that you can draw on from AWS or varying vendors and how much extra that you have to kind of add to that to make it suit in those environments?

Josh Robb:
That’s a great question. So we started working on this in ’23. So there wasn’t— there were a couple of products in the market, but only a few. And we didn’t like the data posture of any of those products. All of them were capturing that data and using it for training. And we felt from a duty of care perspective, that was really unethical, frankly, and also probably illegal under the Privacy Act. So we decided to build our own. And so we’re using Bedrock and Anthropic’s models for inference, and we’re using AWS Transcribe for transcription at the moment.

Josh Robb:
And so those off-the-shelf things work really well. And it’s not just transcription. So we transcribe a consult with consent, but then we summarise that into a medical note.

Paul Spain:
Okay.

Josh Robb:
And that’s the real secret sauce. So actually, our Chief Medical Officer, Dr. Graham Denyer, is the main prompt engineer on this. And he’s— because he’s got a clinical brain, he actually has a computer science background as well, which is super helpful for us.

Paul Spain:
That’s great.

Josh Robb:
Information systems background. So he’s been really invaluable in terms of helping really get the output clinically where we need it to be. But we also capture a ton of metadata off that that allows us to measure the quality, because we feel like there’s a big difference between an individual doctor deciding to use a product in their consult room and us providing this to hundreds of staff all at the same time. And so we feel like we’ve got a duty of care to be able to measure the quality and the impact of that. So we collect a whole bunch of metadata off this to go, how is this performing? Is this performing differently for people from Asia than it is from Māori or Pasifika compared to Pākehā? You know, how is it going for people with different deprivations or different— in different parts of the country, even different practices? You can have a noisy AC in a consult room that’s messing up transcript quality or not producing as high-quality notes. So we’re really interested in making sure that we’re not just putting technology in there, but we’ve got good governance and systems and quality control around that, and that we can make improvements to it over time. So that’s kind of one dimension of it. But the other thing that we found quite early is that a transcript by itself isn’t enough to produce a high-quality medical record.

Josh Robb:
And in fact, what we started seeing when we first started kind of testing this in production was quite a lot of hallucination. You know, the And the model would spit out, oh, you know, this patient is a male who’s 45. And it’s like, they absolutely are not.

Paul Spain:
Like—

Josh Robb:
Wow. And so in response to that, we started grounding the context for the models in your medical record. Because we’ve built it ourselves, we can integrate it into our electronic health record. And so the agents have access to, with very good, let me be clear, very good security controls, they have access to some demographic information about you. They know your name, your gender, your age, your ethnicity, But they also know your medication list, they know your long-term conditions, and they can look at recent encounters. And that becomes really important because if you’re seeing a doctor about something transactional, it’s kind of— the story’s got a beginning, a middle, and an end in a single consult. But often when you’re working on a long-term condition or a condition which requires some investigation, follow-up visits don’t have the whole story in them. Like, the transcript by itself is not sufficient to understand what’s happened because there’s shared context between me as the doctor and you as the patient that isn’t ever articulated because we talked about it last time or the time before.

Paul Spain:
Gotcha.

Josh Robb:
And so giving the agents, being able to ground them in the history of your medical record means that the quality of the notes they produce is phenomenally better than it would be just from a transcript. So that’s kind of been very, very powerful for us and is allowing us to do some really high-quality stuff. And certainly our staff that use other products and other environments, because lots of people work part-time in different Yeah. Places report that what they see from us is much, much higher quality output. Our team can be really responsive to their needs. So, you know, maybe 6 weeks ago, one of our GPs was saying, hey, look, this isn’t working very well for procedural medicine. She does a lot of women’s health, so she’s helping put in long-acting reversible contraception, and that’s procedure. And so you’re not, doing a consult with a patient, you’re trying to document what happened after the fact.

Josh Robb:
And so the team went away, had a bit of a think about it, shipped a feature that lets her kind of record procedural type stuff really, really easily by kind of dictating to the machine what’s happening and then having that turned into a consistent medical note. So that’s really cool. We are also using it for our nursing, for a bunch of things. We are actually using it for that medical certificate service. And that’s cool ’cause that gives them really nice safety netting in terms of documenting the phone call they’re having with you, making sure they’ve checked their red flags, lets us do QA on that. We’re using it with our mental health workforce and also producing some structured information out of mental health consults. Like one of the really difficult things with mental health is how do you measure how effective a mental health intervention was? Outcome measurement is hard. And the best indicator for that is actually the client’s reported confidence in the plan you’ve developed together.

Paul Spain:
Yes.

Josh Robb:
So like it’s quite common for a psychologist or a counsellor to say, okay, we’ve talked about this thing, anxiety, whatever it is, we’ve given some strategies and some techniques, What is your confidence in the plan from 1 to 10? And you might say 7 or 2 or whatever it is. That’s probably the best outcome indicator that we know how to capture. And we can extract that from the structured data that we produce without having to have someone go and fill in a form.

Paul Spain:
Fantastic.

Josh Robb:
Which is really cool.

Paul Spain:
That’s super helpful.

Josh Robb:
So we’re kind of using this unstructured conversational stuff to start producing structured measurements and indicators, which is super powerful and lets us kind of just look at the quality of the services we’re providing work out where we need to make interventions and, and think about how we redesign things for our, for our patients and for our staff.

Paul Spain:
Yeah, brilliant. Well, thank you so much, Josh. , lots packed in there. We’ve probably burned up a little bit too much of your time, so—

Josh Robb:
No, it’s been a pleasure.

Paul Spain:
Thank you for being, uh, so flexible. , yeah, look forward to keeping in touch and, and chatting again in the not too distant future around how things continue progressing.

Josh Robb:
Thanks so much for having me, it’s been great.

Paul Spain:
Good stuff, well thanks everyone for listening in and of course a big thank you to our show partners, PwC, Fortinet, Workday, One New Zealand, 2degrees Spark and Gorilla Technology. That’s us for this week, we’ll catch you again on another episode next week. See ya.