How two brothers turned their father’s paperwork nightmare into a solution for healthcare’s hidden revenue crisis

You and Josh both witnessed your father’s struggles. What was the moment you realised this was more than just one doctor’s problem?

Joel Freiberg (JF): My dad is a lung specialist in Sydney, who’s been in the industry for 30 years. He sees 30 to 40 patients a day, doing full consultations—15 to 20 minutes each. That’s 10 to 12 hours every day. He hasn’t really had a lunch break in 20 years, and this is common for most doctors. 

But it’s legally required, and important for patient wellbeing, that there’s robust documentation of everything discussed [in a consultation]. What medications are you on, how have your symptoms changed, what were your test results, what’s the treatment plan. When the consultation ends, the patient can forget what they were told. If something goes wrong with the patient, the medico-legal people will say one thing [to the doctor]: show me your documentation.

So after those 12 hours of consultations, my dad would have 30 to 40 detailed reports to read through and verify. Previously, he’d dictate quick audio summaries based on memory, which has room for error, and had two full-time typists transcribing them. Some doctors would type while speaking to patients, which means their attention is split and they might miss something crucial.

Josh and I started speaking to our dad and about five of his colleagues—different specialists, top cardiologists, oncologists in Australia—to understand the pain points of their day-to-day work. We did this for six months, building the technology for their practices and then another six months to get the first version to a point where it was reliable, before we started selling it to other people. We took our time at the start because accuracy, quality, simplicity and personalisation were all things we needed to get better at before we expanded the business.

Read more: Why running a startup is like playing 3-D chess, says Reach52’s Edward Booty

Medow’s gone beyond clinics to launch a product for hospitals. What made you realise the documentation problem was even worse there?

JF: Hospitals have an even bigger gap. A doctor visits patients in their beds, often for just 30 seconds, because they’re so time-poor. They’ll quickly check how you’re feeling, look at the chart, blurt out instructions and rush to the next room. The nursing staff will be responsible for executing their instruction for the next 24 hours. If they’re not there, there’s a real documentation gap. But what’s really interesting is how [bad documentation] has a big impact on a hospital’s revenue. 

Here’s how it works: in private hospitals, the insurance company pays the bill. Your bill depends on everything that happened during your stay—if you had an infection, needed dialysis, every treatment has a code that equals a cost. But if a doctor doesn’t document that you needed dialysis, the insurance company won’t pay for it. That might cost the hospital $3,000. Multiply that across thousands of patients, and depending on the hospital size, you’re talking about hundreds of millions in lost revenue.

There’s a real problem in Australia right now with private hospital financial viability. If you’re a small private hospital, that amount of revenue loss could shut you down. And when you have more hospitals closing down, this reduces people’s access to care, which has long-term effects.

That’s an extraordinary figure. Why is this only being addressed now?

JF: The pain points have always been there, but the technology to solve them reliably enough simply didn’t exist. Voice technology has been around—think Amazon Echo or Siri—but without artificial intelligence (AI) behind it, it couldn’t actually solve the task. It would transcribe, but couldn't understand context or produce quality output. With the breakthroughs in AI, we now have the tools to build solutions at the level healthcare requires. 

Read more: Can a Furby or Tamagotchi cure loneliness? Inside the billion-dollar smart and AI toy boom

How does Medow train its AI with deep and specific healthcare knowledge?

JF: We use publicly available data, such as medical journals, but mostly use the knowledge of the specialist doctors who work closely with our AI team. We spend a lot of time with the doctors to get their feedback on what’s missing, what’s wrong, and iterating together to the point where the system can learn on its own from the parts [of the transcript] that the doctors are editing.

We’re now processing around 5,000 consultations per day between doctors and patients, all anonymised. By design, AI can have errors, so our job is to refine the model, putting in automatic checks. That’s why it takes 15 to 20 seconds to generate reports—our AI is producing, then checking, rechecking, referencing where something was said in the transcript. We need 99.9 per cent accuracy, so doctors only have to fix 0.1 per cent.

What’s been the most common concern or reaction from doctors about adopting a product like Medow?

JF: Scepticism was maybe one thing. Doctors will be like, “You say this, but can it really do it?” So we demonstrate it. We do a mock consultation—if it’s a cardiologist, we simulate a patient with chest pain discussing test results. We actively take the conversation in various directions, like a real consultation. The doctor sees the technology working in real-time, from start to finish.

Then we suggest they try it with one or two patients a day, see how it works. Every clinic gets a dedicated account manager who checks in, supports them and helps customise documentation to their style. That high-touch service is critical. These are very busy, not necessarily tech-savvy clinics. We’ve learned that it can take time to handhold and support, and that also gives us constant feedback to improve the product.

What about data privacy? That must be a significant concern.

JF: Absolutely critical. Before Medow, I built a health insurance company in Australia, so I’ve dealt with these exact compliance requirements—patient health data plus financial regulation.

First, data residency is a must-have. Doctors, clinics and hospitals own their own data. All Singapore data stays in Singapore. All Australian data stays in Australia. We partner with Microsoft for cloud infrastructure—they work with the biggest healthcare institutions to secure hospital systems. Our AI models are stored locally.

Second, anonymisation. When we send data to the AI model, it never sees patient names. The AI sees “patient first name” and “patient last name” as variables, not actual identities. The doctor sees your name, but when we’re capturing information, summarising, creating documents—all those steps in that 20-second wait—there are never names in the process.

Third, we don’t diagnose. We don’t infer diagnosis. Our job is to capture what was said, organise information and ensure accuracy. If we capture a diagnosis, it’s only if the doctor specifically stated it. Our AI looks for “did the doctor say you have pneumonia?”, not “did the patient say they think they have pneumonia?” By design, the doctor must validate every report before it enters the patient file. We’re an assistant, not a replacement.

Read more: The app that can turn your phone or laptop into a pulse monitor

You chose Singapore as your first international market. Why?

JF: Three reasons—two professional, one personal. Professionally, Singapore is a medical tourism hub. If you can help doctors here, you’re helping patients from all over Southeast Asia. 

Second, it’s a hard problem. Singaporeans are discerning; they appreciate high-quality [care]. Doctors here are passionate about their reputation and patient care. The government and the Ministry of Health are tightly integrated. If we can solve for a market like this, we can solve for anywhere.

Also, we focus on specialists, so we’re going after the hardest problems. If you solve for specialists, you can move to less hard problems.

Personally, I lived here for a year building Pelago, a travel experiences booking platform, with Singapore Airlines from 2018 to 2019. I worked with amazing people, built strong networks. Our country lead, Clement Tan, is a former colleague from that project. I also grew up coming to Singapore with my family—we’d always transit here on the way to everywhere else in the world.

What were the biggest surprises when you entered the Singapore market?

JF: The insurance dynamics. We’ve learned that the burden has shifted from patients to doctors. There are cheaper insurance premiums available, but what patients don’t understand in the fine print is that insurers can challenge everything that was said. Yesterday, I was at the clinic of a thoracic surgeon, and he’s getting emails and WhatsApp messages from insurers asking, “Why did you order this MRI? Did [the patient] really need it?”

That was an interesting dynamic where evidence is so important. And there can be real gaps, especially the busier the doctor gets, because the less time they have to ensure that every decision they make is documented. If they don’t have robust documentation with the “why” behind every decision, insurers can refuse to pay. 

Singapore is multilingual. How does Medow handle that complexity?

JF: We support hundreds of languages—Mandarin, Bahasa Melayu, even Singlish. We partner with language translation providers who’ve built AI models just for this and integrated theirs with ours. 

What’s powerful is that a consultation can be done in Bahasa or Mandarin, and the reports that are produced will come out in English. Our AI can also understand context. If there’s a missing word in a sentence, the AI can work out what it should be based on the surrounding context. That’s what makes it work with Singlish or dialects.

Read more: Illuminating Asia’s Stanley Tay launches HeatSig, an AI-powered light that can save lives

Where do you see Medow in five years?

JF: We’re focused on documentation now, but there are endless opportunities to solve problems in healthcare with AI. The next area we’re excited about is billing. It’s also very manual, with a lot of room for human error.

Final question: what is it like working with a sibling?

JF: The best thing is actually getting to work together. I’ve lived in Singapore, London and worked in different parts of the world. We didn’t spend much time together for 10 years. Now we’re working together, travelling together, building this [company together].

We’re very different yet complementary. I’m more product and technology-focused, while Josh is market-oriented—he loves meeting customers, partners, understanding their stories, learning from them and selling the product. He was setting up Singapore for months before I arrived. We have different core responsibilities, but we’re part of everything.

It is, however, harder to give constructive feedback to family. We’re working in healthcare, where everything needs to be of the highest quality. If something isn’t good enough, telling your brother can lead to squabbles because, in our minds, we remember being five years old, running around hotels, fighting. That history is there. But the benefits outweigh the costs by a lot.

NOW READ

The “tough guy” myth is affecting men’s health—and how this Filipino entrepreneur wants it gone

Forget charity: how Jollibee’s foundation is building farmers’ dignity instead

Building obsession: Steven Bartlett and Hong Kong’s rising tech founders on people, purpose and the power behind innovation