How We Flipped the Playbook to Build Australia's Fastest AI Company

Hello, HubSpot Grow. I'm Tom. Thanks so much for coming to this session. It's going to be exciting. Everyone in this room sells something, probably. And you try to get someone's attention, you earn their trust, and then we turn that into revenue. Now imagine, like me, your customer is an overworked doctor. They don't really take sales calls, they don't really respond to Covid outreach, and they usually barely have time to eat lunch. And that was our market, that was the one we chose, where I'd worked and we had no money to advertise and we had to figure out how to get cold customers. And so in that context, it kind of feels surreal to say that Heidi is the fastest growing company in Blackbird's venture portfolio, probably one of the fastest in Australian history. It's kind of unbelievable just how many patient visits we're in every single week. So more than 2 million, something like more than 100 countries. And we have staff in more than 10 countries now, and we recently crossed 50 million USD. We've managed to achieve all this in about two years. And I would love to tell you that this was my genius master plan, all coming together. Everything worked perfectly. But I think the reality is it's actually a lot more complex than that. We actually flipped the traditional enterprise playbook, mostly because the old playbook of going top down nearly killed the company. It was very close. But to explain how we actually got here, I'm going to take you back to where this all started. So this is me in my scrubs. I was working as a vascular surgical registrar. I was in Melbourne, actually in the country when that photo was taken. And I got to see a front row seat to how the healthcare system was overwhelmed, overworked. I felt that as a clinician and I also saw what could happen to patients. And I realized very quickly that healthcare is, you know, not exactly a perfect go to market laboratory. You know, you're trying to build something for very busy, highly educated, deeply skeptical, overloaded clinicians. And they are trying to get them to adopt something new in an already busy, frantic working day. And the only way you can get them to use something is if you've built something overwhelmingly valuable. And so to give you an idea of what it's like in surgery, you would have 50 patients on your list, you'd have a theater waiting, maybe patients in the ED backing up. And if you're lucky, you would have five minutes with each patient. And the patient outcomes, like what actually happened was enormously dependent on me or whoever the Doctor was in the room, whoever's at the wheel. And it's a pretty confronting, taxing way to work. You're constantly on the front line, you are the capacity constraint. And I realized that this capacity problem wasn't exactly a spreadsheet problem. It was about human bandwidth. We could only do so much. And I didn't leave medicine because it was meaningless. I would have been perfectly adequate vascular surgeon, I think, but I could only ever see one patient at a time. And to me that wasn't enough. I didn't want to be in a broken system, want to try to fix it. And so I had luckily had some experience in tech before medicine. And so I was, I thought that the only way to fix this at a big enough scale was through technology. So I cold emailed, I think it was Nick Crocker, one of the Blackbird partners at the time at the hospital. And I remember he took my email and then organized a big pitch like Robin Denholm and all the Blackbird ic. And I'd been on call overnight and I couldn't get the day off. So I literally snuck into a cafe in blood soaked scrubs because I'd been operating that night to pitch the Blackbird IC at the time. We had like a toy product, didn't really do anything. It was a cool demo of what it could maybe do. I had a, you know, outrageous five year plan of how it would go up into the right, as all startups do and not much else really. And I told them and this part I was sure that every doctor and probably every patient would use AI in the end to do healthcare. Like I wasn't exactly sure what to build, but I believed in that AI care partner that we could build. And so I think the thing that I've learned and the thing that you have to believe is that if you build something with real utility, like something that's really useful, people will use it even if they're not officially allowed to use it. Cause I used to do that at the hospital all the time. Like if I thought something was good for my patients, I would just use it. Sometimes get in trouble, but I would use it. And I think Blackbird believed in that too. So that was why we ultimately went down a product LED model. But the part of the kind of founding story that usually gets kind of skipped is what happened in the first few years where we got this entirely wrong. So initially we actually built this very complex AI infused telehealth platform. This was during COVID so you can imagine there was like a Biggest market distortion healthcare has probably ever seen. And we followed the traditional SaaS playbook. So we sold it top down. I would go pitch a practice manager or a private equity firm and say, heidi can transform your business, you know, you should use it. And it kind of worked. Like we won lots of contracts, you know, popped the champagne, got excited, like the hypothetical revenue was going up. And because we were selling to C suite, we kind of ended up dictating our product towards what they asked us for. Because we were trying to win deals. That was what we were ultimately optimizing for. Inevitably, we built exactly what was required to sell to the buyer. We added compliance dashboards and reporting metrics and administrative gates and all these different things. And when we finally got to doctors, it wasn't really what they wanted. We were basically asking them to add about 14, 15 extra clicks to achieve some sort of C suite objective. And it was changing their entire workflow because this platform needed the patients, the practice managers, the receptionists, the doctors to change the way that they worked. And so actually what happened was the doctors just hated it, they didn't use it and we became shelfware. And so by the end of 2023, remember we fundraised at the beginning of 2021, we were pretty much out of money, we hadn't landed the plane, so it was tough. But I had to fire about 20 people who had taken a break in their careers, trusted me, believed my promises, and we had a tiny team left and we had about six months of Runway for this tiny team and we had to kind of figure it out or just shut it down. And I'd go back to, you know, operating and writing notes. And so what I learned was that when the buyer and the user are different people, you can win the sale. Like you can say the right things, you can convince the C suite person like to, you know, sign the contract, but ultimately you will just lose once the product gets to the end users and they don't use it. And so then what happened, as you may or may not remember, was ChatGPT launched. And to put it lightly, we were just entirely cooked because we had spent all our money and energy building our own machine learning models. Like I was a maths major, machine learning person before medicine, so I thought that our company was building the models. And then I wake up one morning and GPT4 is better at everything than what we had built, like everything, you know, and so everything we built was basically obsolete. And so not only were we now pivoting, but we literally had to start Again, basically it's like everything we built, worthless. What are we going to do? But it did kind of create an opening for a new product. And the thing that we're starting to notice was all these doctors who had been so wary of us. I've never transcribed my patient, I would never use AI. They started asking us different questions, like could I use AI to solve this problem? Some of the more very excited doctors would say, can AI run my whole practice? That would be the dream. And even more simply, can it write my notes? And so we swallowed our pride and just stripped everything back. Like simple to be smart, just do one job, transcribe the visit. So like listen in the background, turn the conversation into words, use an LLM not our own. Just eat it and use the best available, write the clinical note and that's it. No grand platform story, no like changing the world. Let's just write the notes, create immediate value for the doctors, do the job that they want. But the problem was like, how are we going to get people to use it? How do we get our new product into market? This wasn't a unique insight. There were a lot of companies that had a similar idea. You probably all remember the litany of different note and transcribers that you've all seen. And this was probably the biggest, riskiest decision that we made in the whole company. We made it free. Now all the CFOs in the room, hopefully not my CFO will lead. My co founder probably are having and had a heart attack, will lead still with us. And he was basically, he just couldn't, I guess, justify how such a compute heavy task could be made for free. Like how is this business going to work? He's just going to burn all the money and die. And it was terrifying. But it felt like we had learned enough from how doctors use these tools that we thought it was the right decision. To give you an idea of how terrifying, it would cost us about a dollar to generate a note, give or take. So some doctors would do about 30 visits a day. So you know, we are paying $30 a day for a doctor to use Heidi with no revenue. Most people thought we were insane, but we didn't have any marketing budget, we had no money on ads, we didn't go to conferences, we had no outbound SEOs, no sales team. It was this, you know, 10 to 15 people just usually. And we realized early on after making it free that we didn't really need to buy ads, we needed to buy distribution. Like we had to find a way to get people using it. And we took inspiration from the canvas, Notions, Figmas, all the products that we loved. And we decided to treat that compute cost or the server cost as our marketing budget. Like, we're all in. There was no other option. If we're going to do this, we can't really afford anything except paying for the service. And we thought that $30 a day on the face of it maybe seems too much, but actually if you view this as a customer acquisition cost, maybe it's not that bad. Like we can find a way to reduce the cost and hopefully they will pat us on the back in some other way. And so we went from traditional enterprise, you know, 90% gross margins to consumer app overnight. We had to just remove every single piece of friction on the product. So we just had to make it super, super, super slick and, and easy and just deliver value. Like no credit cards, no sales call, no IT approvals, just make it incredibly easy, sign up in seconds, kind of like hack them into experiencing the experience with a patient even though they didn't have a patient in front of them. And they just got the magic straight away. And so pretty much immediately, actually this was like February 2024, by the time we went this direction. We just saw this immediate shift in behavior. You give a burnt out clinician like me something that will save them an hour a day minimum, sometimes two hours a day or more. And they definitely don't keep it to themselves. Like they tell everyone, they tell the doctor in the next room, they tell all their patients. Some of those patients are clinicians as well. And it creates this incredible viral loop. So we found that pretty much if one doctor did about seven sessions with the new version of Heidi, their retention was about 90% every week after that. So they just like, they're stuck. They would never stop using it. It became an indispensable part of their daily workflow. And that was when the commercial part of the business started to work. So once you earn their trust, then usually in like week two, after seeing the notes generate, the initial blush of how exciting it is wears off and they're like, I really wish Heidi would write like this way. It's a bit annoying, isn't it, how it writes this way. God. And they just forget what it was like before. And we built product that they could pay for to customize Heidi in the way they liked, integrate their medical record, invite their colleagues, et cetera, et cetera, and that was what they would pay for. So classic product led growth strategy. The other key thing was all of the effort we made in reducing friction in the onboarding helped speed up this cycle. So if someone told a colleague, they could also just sign up straight away. So there was nothing that would slow down the growth of the product because they could just sign up, start using it, and it grew from there. So making it free in the end actually was the best way for us to generate leads for sales. It wasn't just about giving the product away for free. And intentional or not, Australia was actually the absolute best place to do this in healthcare, because in healthcare, this model never works. Like, there's not a single product I think, like maybe one company I can think of that has done product led growth in a way that worked. And it was because Aussie GPs are usually independent subcontractors, so they actually do have a choice. They're not employed by huge health systems. So if they find something useful, sort of more or less, they can just buy it and use it. And so our free tier was an incredible distribution strategy in Australia. It did work in Canada and the UK as well, and it matched the market super well. And basically we also found that if the product is strong enough, the users will just fight for us. Like they will tear it into their enterprise. We had people saying that they wouldn't work if the hospital stopped them, allowing them using Heidi. So pretty powerful. We have many advocates, but Dr. Catherine Scallon's my favorite. Catherine's a developmental pediatrician. She works in Queensland. She started using Heidi for free. She was drowning in documentation and in pediatrics, it's especially hard because you can't write notes while you're engaging the kids. So it's like this burden of work that would happen every night for her. And so she can now actually look at the kids and the families and be a better doctor with Heidi doing the rest. And she just loved it. She loved the product. And so she loved it so much that she took it to the CEO of Children's Health Queensland Hospital, where she works, and she won an innovation grant completely off her own back, like, we didn't even know she'd done this. And she spearheaded the biggest digital AI scribe project in all of Australia at the time, across the entire hospital, for all the specialists, all the allied health, all the nurses. And by the time we had a sales team, at the time, it was just me. And today, when our enterprise sales teams engages hospital leadership, we're not really cold pitching anymore. We're just saying, hey, your insert clinician names already. Love this. We've proven it works. Let's explain to you what the benefit is for you, because we know you don't pay the doctors for overtime anyway, so what should you care about? So that would be billing, coding, other things. So the value prop for the higher up still has to be there. Like there has to be something, but it's not the same as what the doctors get. And we talk about enterprise compliance and how we roll it out and integrate in their systems and all the rest of it. And we have Catherines all around the world that help us build those relationships with enterprises in a way that we couldn't without that product led motion. And so the truth is that our best AES, like our best account execs, will never ever be as persuasive as Catherine telling their CEO what to buy. And so since we're at a HubSpot conference, I can tell you that this entire motion relies heavily on our CRM. So I think the shift was we realized that the CRM wasn't for spamming cold leads. It was a way for us to kind of listen to organic signals from the product. So when a clinic or a hospital starts to light up, there's some organic usage, we see an account start bubbling up. We can use that data to tell exactly when our team should step in. Like when's the right moment? Who are the right key opinion leaders that seem to be using this? The Prof. Of cardiology just signed up. Let's send them an email. And it's a completely different approach to normal enterprise sales. And so this is the power of a bottoms up motion. When you own the hearts and minds, like really, they love your product, the end users, the rest of the market just sort of bends to your will. Now, our product led growth promotion came off the back of our scribe. But for us, saving a doctor an hour or two or just writing their notes was just the beginning. Our goal was always to build an AI care partner, more complete AI that would go across the whole AI care journey, both patient, doctor, everything in between, and can we absorb the entire friction of practicing amazing healthcare? So here's how it works. This is what clinical life kind of looks like. So if your doctor isn't using Heidi, she usually would arrive at say 7am, spend maybe half an hour, an hour reading all the results that came through, marking off letters, sending off tasks that have to be done before the day starts, and then seeing about 25 patients back to back to back, and then maybe kept up during the day, but usually didn't. So there's about two hours of work waiting at the end of the day. You try to do that as fast as you can get home for dinner and, and then you usually end up leaving at about seven and you're exhausted and you get home at eight o'. Clock. And so if your doctor is using Heidi, what our promise is is that Heidi can actually already handle a lot of the routine communication around the visit before the clinic even opens. 24, 7, you can call Heidi anytime. When she walks into the room, she's wearing a Heidi remote. 21 grams lightweight, captures all the high fidelity audio. She doesn't even need to remember to sign in, she just presses the button and she's immediately transcribing and it looks like nothing, but it's a little device and gives their doctors their lives back. And she starts talking to you, you've come to see her, she's doing her visits. Scribe in the background is automatically creating the best note. Maybe you mentioned some rare symptom or it triggers a memory of something that may be going on that she can't quite remember. If she had the time, she could definitely look up all the research and the evidence, but she doesn't have the time. So evidence pulls the latest global medical research in seconds right there in the room in the background and your doctor's still looking you in the eye, actually talks to you, gets the management investigation right and goes home at 5pm because she did the notes the whole way through as she went. I think ultimately there's one final reason why our PLG playbook worked and it's relevant to everyone here who's working in a company or building a company. I think people often talk about building in Australia as a disadvantage, but for us the constraints of being here was actually a real superpower for us because Australia is smaller than some other markets so we could actually be really close to our users. And because Australia is never big enough to hide in as a company, especially a venture backed company, we were forced to think about how we would go globally. Like how would we make this work overseas? We didn't pick countries from a spreadsheet or do desktop research. We just followed where our users were and we basically just looked for places where we already had Catherines, we already had believers, people begging us and asking us to come come to where they were. So we sent in regional GMs, people from the initial team. We spun up legal and compliance and local processing and everything we needed to do and basically layered on top B2B sales in these markets where demand was already kind of starting to catch fire. And so I think Being in Australia demands clarity. You know, you have to be focused. You cannot raise as much money as your competitors. You have to build a better thing. You have to like, it's kind of like David and Goliath, whereas we're perpetually David, but just like with better weapons or something. And you need to be ambitious. Like you need to make a product that can work in the initial innings here with these demanding users, you get direct feedback and then you take it to the world. And so I want to leave you with three things today. The first one is, and this is the most important lesson, the thing that nearly killed Heidi. Don't confuse the buyer with the user. And especially, I know it's nice to make quota, but like it sucks when that churns and the business you're working for shuts down. So make sure that you're nailing both. Like the product has to work for the end user, ensure the buyer needs to be happy. But you have to get both right more than ever. With AI, the conventional wisdom of doing top down sales doesn't really apply because AI more than anyone is a very personal thing, as you all probably experienced. How do you actually get adoption and usage of these tools? The products have to be incredible. The second thing is using constraint as an advantage. It doesn't just have to be Australia, it could be any constraint that you, you have because it forces you to make trade offs and make more incisive strategic bets which I think help clarify what you're doing. Like what are you focused on? You get faster feedback. You focus on the users that are working rather than the things that aren't working or some contract signed or some other surrogate metric that's not the real thing. And then you innovate constantly to try to escape your constraint. And the third thing, go to market strategy has to be in service of some mission or something. Like for us, yes, PLG is nice. It's really cool that we got this to work. It's differentiated in healthcare in a cool way, but it's because of the mission that we always set out to do. Like why I stopped my surgical training. So product led growth is not our mission. Bottom up's adoption is not our mission. The mission is building an AI care partner that can actually solve the healthcare capacity problem. I can still see it in my mind, like in 5, 10 years we will all be able to get some form of ambulatory well medicine from something like Heidi. And then when shit hits the fan and you're really sick and you need to be in a bed or you broke your leg or whatever, you go see a doctor who's also using Heidi. And so that's why we call it a care partner. It's a partner in your journey and your doctor's journey. It's not trying to steal patients away, it's not trying to change the way doctors practice. It's just extending their reach and increasing their capacity. And we can do this at a global scale. So we can do this for individual doctors, allied health nurses, physios, but also for whole health systems that we now service. Everything else, like the tactics, the how for us is just kind of plumbing of how to get to the patient, basically. If that resonates with you, everyone in the room, we are hiring, please join us, especially if you know your way around a HubSpot pipeline and we don't need to wait for anyone else. Like we are building the future of healthcare for Australians and the world. I think we're going to be the most important healthcare AI business in the whole world and we have the talent in Australia to do it. And when you go back to work tomorrow, remember that the real question is not who signs the contract, the real question is who loves your product like so much that they would fight to keep it. You've got to try and find those users. Thank you so much.