00:05:13 · Matt
[00:05:13] S0: Yeah, so I unfortunately had a similar experience with my own grandparents watching them drift away, seeing them present but not really present. That got me very interested in this space. And so I think it was around 2014 I was working at Bridgewater, which is this very large hedge fund. And my nights and weekends, I was reading audiology textbooks. And I was lucky to have my best friend. His mother is a professor of audiology and a really prominent influential professor of audiology. And she was just very encouraging of my interest. And after a few months, I got very discouraged because I realized that building a better hearing aid for my grandparents was, it seemed like too hard of a problem. The more I learned about the space, the more intimidated I got. And just to give you a primer, age related hearing loss is sensorineural. So it's not just a problem in the ears. It's as much a problem of the brain as it is of the ears. And so when I came to appreciate that it's not enough just to make sounds louder, You have to differentiate between sounds that are important and unimportant. This in 2014 was very mis... Misaligned with the capabilities of digital signal processing techniques that powered hearing aids. So when I appreciated just how intractable the problem was, I gave up. I just put it on the shelf, and I didn't want to work on hearing aids unless I really thought I had something to contribute. But that process of learning about hearing aids twelve years ago got me very excited about spending my career on medical devices. And so I was very fortunate to find a job at an early startup company called Butterfly Network founded by this guy named Jonathan Rothberg, who was a pioneer of genome sequencing on a semiconductor platform. This is a big deal innovation. If you've ever known anybody who had cancer and beat it, it has a lot to do with this innovation of sequencing the genome on a semiconductor platform. It's called next generation genome sequencing, and it changed the world. This guy had this playbook for taking a conventional industry and disrupting it with custom chips. And he had done this very successfully in sequencing, and he wanted to do it in medical imaging. And I heard about this. I was like, this is what I got to spend my time on. And so it took me about a year. I just begged, begged for a job.
00:07:59 · Matt
I became sort of an expert in medical imaging technologies, eventually joined as a product manager at that job. And for six years, my colleagues and I, many of whom are my colleagues again, my co founders again, devoted ourselves to this mission of building a better medical imaging system. We thought of it as a visual stethoscope. And so it's a handheld medical imaging system that itself is a custom chip, and it connects to an iPhone, and it's about 100 times less expensive than a conventional medical imaging system. And we're very lucky that it became the best selling medical imaging system of all time, and we were able to take the company public a few years ago. After that experience, my colleagues and I had developed a bit of a playbook ourselves for how to do stuff like this. And it was after that experience and the confidence it gave us, but also the network of engineers, window into what was possible with AI, that encouraged us, me and my co founders, to take another look at hearing aids. And this time, we already had experience as a startup company disrupting a major industry, building a radically new kind of custom chip. Deep learning had emerged. The world was quite different. And so I felt, and I will say, sadly, the next generation in my family now was struggling with conventional hearing aids, and I was shocked again to see that the technology had barely progressed. So this was the call to action and we felt like, all right, if we don't do this, who's going to do it?
00:09:46 · Adam Miller
[00:09:46] S2: When you consider hearing aids, they do an incredible thing for the world. They help people hear better. But the reality is, is that traditional hearing aids aren't that great. They got to experience it in this episode. And what Fortell are doing is that they're putting incredibly powerful AI into hearing aids to make them 10x better. And they've meant that people can hear so much clearer and just have a better overall living experience. Now, when you relate that to Notion, Notion's always been a great tool. We've used it to run literally everything in our business. But quite recently, they came out with Notion AI. It's AI that sits on top of all of Notion, allowing you to search for any docs, allowing it to do work for you. You can spin up little agents in your Notion. And so they've gotten a product that has been great and made it even better by adding AI on top. And the really sweet thing is, is that we're putting everything into Notion. We're giving it more data points about us over time. And what that means is that Notion AI gets even better over time as well. It learns more about us, it has better context, and it can make better decisions for us. For anyone out there that's not already using Notion, we've got a special offer for you, three months of Notion business for free. If you head to ntn.sosachinandatom, you can get it there. So the call to action and the problem with conventional hearing aids sounds very obvious and explicit there. But it seems like for your company, AI was the big unlock. Can you sort of run me through how you got this insight that AI might be the unlock that's needed for hearing aids and how you developed the conviction to get to the point where you're like, wow, I can develop a whole company based off this.
00:11:25 · Matt
[00:11:25] S0: Yeah, so it actually goes back to COVID. So the context here is I was primed from my previous passion investigation, maybe you could call it obsession for making better hearing aids. I was primed. I understood the problem, which is that people with a hearing loss don't just need loudness. They need selective loudness on the important stuff, And otherwise, they need quiet. If you could take grandpa to a really quiet room and put in his conventional hearing aids, that's going give him the best possible chance of staying connected to conversation and the people he loves. So I knew that this was the issue. It had been on the shelf. I'm doing this company with... As an early employee called Butterfly, where we're doing our own custom chips, our own AI, our own manufacturing, full stack medical device project, and COVID hits. And all of a sudden, we're thrust behind our laptop screens at home. And I don't know if you guys remember that early COVID, background noise in video chats was a major problem. You could hear when people were typing, peep... You know, people were always kind of shushing their kids. You have to turn off your air conditioning units, telling people to mute. It was a nuisance. And then gradually, the background noise just got quieter and quieter. And it dawned on me one day in a meeting, I was watching my colleague type and I couldn't hear him. And I had this kind of moment. It was like, woah, Zoom and others like it, they are doing what was impossible. Right? They're they're differentiating between sounds on the basis of what they are, not what frequencies comprise them. And so I had this like kind of realization that this technology that had been deployed into telephony and video conferencing could be a major, major game changer for hearing aids. The question was, could could it work? There was a proof of principle in video chats, but in in a video chat, the problem is way easier in kind of every dimension. You have much easier input signal to noise ratios because in video chats, you talk directly at a microphone that's positioned to pick up your voice. You're very close to that microphone. So you have very easy incoming signal to noise ratio. You have this very large computer. Even if it's your phone, your iPhone is a very large computer relative to a hearing aid. Hearing aids are tiny, and they need to last all day, right?
00:14:02 · Matt
So in terms of your ability to access compute and power, you are super constrained as a hearing aid. You have more challenging signal to noise ratios, and maybe the most challenging and most pernicious of all these constraints is the most subtle, it's latency. A hearing aid has to do its job outrageously fast in just a few milliseconds, or else the wearer will notice the delay and the world will just be out of sync. You'll see lips moving and it's like a badly dubbed movie. You can't live like that. In telephony and in video chat, you can buffer, you can delay one hundred, two hundred, three hundred milliseconds and just resynchronize the video with that audio, so you have a lot more time for your AI system to select out what's important. It's the reason in film editing and video editing, you can do a pretty good job denoising in any major editor. A hearing aid is the most challenging quadrant of this multidimensional problem. And so when we started the company, the big bet was that if we if we worked on it very aggressively, systematically, and we're open to inventing new tools, like building new chips, then perhaps we could take this proof of principle and squeeze it into a form factor that could benefit people with hearing loss every day.
00:15:25 · Sachin
[00:15:25] S1: Matt, we're going to go into the future of Fortell, and there's so many interesting things we want to dive into there. But before we go there, I'm really curious about an aspect about your personality because a lot of people will see problems around us. I'm sure everyone watching this has seen problems they want to solve in their own lives. What is it about you that made you actually go after this problem and dedicate five years of your life and I'm sure many more in the future to solving this problem? Are there any clues on maybe how you were brought up or in your early life that you'd go out and solve such an impactful problem?