This week I am talking to Punit Soni (@punitsoni), CEO of Suki (@SukiHQ) who are revolutionizing the healthcare space with technology designed to improve the capture of information.
Punit has an interesting background and journey to this point and his perspective on the next big company we have yet to see is on point. As he says
none of the existing companies are even close to scratching the surface. The platform companies of our generation may do a little bit of healthcare, but it’s not primary healthcare businesses
So what should we be looking for and where might this innovation come from – listen in to find out.
We talk about voice or speech recognition in healthcare and revive a quote for me going back many years (2005)
“Speech Recognition has been 2 years away for the last 10 years and its still 2 years away”
We both agree it is not and there has been some incredible progress but the current excitement around “Ambient” tools is not quite as straightforward as it seems and his thoughts are highly relevant to anyone seeking to disrupt healthcare
Listen in to hear Punit’s top 5 ways in which we will bring about change in healthcare – as he says it is not easy, but it is possible
Listen live at 4:00 AM, 12:00 Noon, or 8:00 PM ET, Monday through Friday for the next week at HealthcareNOW Radio. After that, you can listen on demand (See podcast information below.) Join the conversation on Twitter at #TheIncrementalist.
Listen along on HealthcareNowRadio or on SoundCloud
Raw Transcript
Nick van Terheyden
And today I’m delighted to be welcoming punat. Sony. He’s the CEO of Sookie punat. Thanks for joining me today.
Punit Soni
It’s a pleasure to be here. Thanks for having me.
Nick van Terheyden
So if you would, would you share a little bit of your background, it’s unusual in the healthcare setting. And certainly highly relevant to the discussion. So tell us a little bit about your journey to this point in your career.
Punit Soni
Sure, you know, now that I’ve done healthcare for the last few years, I think there’s, it’s almost a little bit of, you know, humor involved in telling my background, because it’s really, really all over the place. I’m an electrical engineer by training. And, you know, I started my career in electronic design automation, so writing software for chip design, and then ended up actually going to Washington to get my business school, business education, and then went over to venture capital, which was awful. And, you know, I was bored out of my mind and decided I should not do it until I actually learn to build something. And so at that point, then the options were either go to a startup or to Google. And it was really still relatively early days of the company. And I joined it in the search team, which is where you want to be if you really want to know the guts of Google, as a product manager, and did search for a bit news books. And then one thing led to the other ended up actually running the mobile apps group product management team. And so early versions of most things like Gmail docs, chat YouTube calendar, came from my team. scarily enough, some of that stuff still looks the same. And then worked in gaming. When Google decided they wanted to work on social initiatives, they asked me to come and run mobile, social. And while I was doing that, I wanted to actually really started thinking about dabbling in hardware. And that’s when we acquired Motorola. And so I was asked to go run software for Motorola, which is probably the most fun job I’ve had in a long time. It was super fun to build devices. And actually, the first device we built called Moto X was the first always on voice device in the world before Siri, Alexa and others even showed up. And so did that. And then then, because I had not done ecommerce, I guess so I ended up actually becoming the Chief Product Officer of Flipkart, which is India’s largest e commerce company. And so now the running joke is, you know, I’ve done electrical design, I’ve done apps, I’ve done search, I’ve done games, I’ve done mobile apps, I have done hardware, I’ve done ecommerce, it’s probably only logical that I’m doing healthcare at this point. So that’s, that’s a little bit of my background, as you said, unusual and all over the place, but the principles of engineering apply almost everywhere. So you can take it from there.
Nick van Terheyden
So I think, you know, good to understand all of that. And obviously, a breadth of sort of technical insights that you bring to the space, you ultimately went into healthcare. But specifically around the voice and speech, elements of this, tell us a little bit about your steps through that and your experience.
Punit Soni
Absolutely. You know, when first, you know, I don’t really have any emotional reason that I decided to do healthcare. It’s a it’s an ambition reason. When I look at various, when I was trying to figure out what to build, I wanted to do something where time would be a superpower. What that means is that if you actually put in time into something, it would demand that kind of time to actually get things done, that actually makes almost all most entrepreneurs will be weary of the space. And healthcare is one of those where if you have the time and the energy and the resources, you can you can create a inherent moat in even starting to do something. And so I wanted to do that. Number two, it is my belief. And I still believe this after five years in that the biggest, most relevant, most interesting technology company of the next generation is going to be in healthcare. And it none of the existing companies are even close to scratching the surface. The platform companies of our generation may do a little bit of healthcare, but it’s not primary health care businesses. So I believe this is going to happen whether it’s going to happen in five years, 10 years or 15 years, I don’t know but I wanted to take a shot at being a part of that journey. And then Number three, it felt like you want to start in a place where the pain was super real and obvious. And I think the one of the biggest public health crisis in this country that few talks about is the absolute burnout that’s happening among our clinicians, all the way from doctors, to nurses, and the fact that at least a part of that can be attributed to super complicated interfaces that we have provided them. And the unbelievable amount of administrative burden, regulatory burden, we have also added on to them. So now if you put all of this together, it looks like a problem. That’s what’s being solved. It looks like a place from where you can generate the kind of structured data that can help you think about building the technology company of the future in healthcare space. But now the question is, what’s the thesis? How will you attack it? And the thesis that I had bars that were mobile was in 2003 2004. Around that time I started thinking about getting into mobile apps is where voice is today. That voice is going to be the next generations user interaction model user interface, it will be pretty casual for people to talk through voice. And so now if you put all of this together, why is this the next generation user interface, doctors are burning out because they have these complex interfaces of the past administrative income and regulatory burden is really multistep complicated data retrieval and input that can be eased with voice, suddenly, you have a space in which we can play and hence, Sookie, which is a digital assistant for doctors.
Nick van Terheyden
So there’s a lot in there, and I want to come back to the voice piece of it in a second. But before I do that, you mentioned something that I think is important that I want to sort of, you know, stick a stake in the ground at this point. And what I heard you say is that the the biggest company of our time, or of the future, let’s say, doesn’t exist, it’s not one of the existing and in healthcare, we know all of these names. In fact, we know, in general, in business, we’ve got these names, but you’re essentially suggesting that the company that is going to be just completely market dominant, doesn’t even exist today. Is that true?
Punit Soni
Yes. And, and the reasoning behind it is pretty logical. If you think about media, you got Netflix, if you think about search, you got Google, you think about social networking, you got better. If you think about e commerce, you got Amazon. But healthcare is even bigger than all of these sectors. Healthcare is 20% of US GDP 5% of any country five to 7% of almost every country’s GDP. Now, somebody can argue with me that that’s actually a lot of small, big businesses. But I believe that one could say the same about a lot of the sector’s I just mentioned before, also, I think the key is to figure out what is the front end or the back end of all of healthcare from a tech perspective. And so far, there’s not a uniform company that’s pulled that together in a way that they can actually build the rest of the stack. So I think, I think there’s going to be a really significant company, at least of the order of these other companies that’s going to be built in healthcare. I don’t think it’s built yet. And I do think that there’s a lot of opportunity to start, including the areas that we’re looking at, which is the front ends of healthcare, tech, and voice.
Nick van Terheyden
Yeah, so I think that’s important. And you know, I can’t disagree with that, although there’s a part of me that goes, it’s really hard to imagine that given the size of some of these organizations, and you know, I’m sure there are individuals that will listen to this and say, well, there’s no way nobody’s ever going to dominate it. But ultimately, that’s the way that commerce works. That’s the way that economics works. And I think importantly, to this conversation, central to this is part of your thinking around voice, which you know, people see as a tool, but you’re seeing it in what I consider quite a different concept of an interface. And I guess we see a little bit of that in Hollywood, but you know, that’s, that’s Hollywood, it doesn’t exist, right. But what you’re suggesting is that it does exist and it could exist and could contribute to both that large organization but also importantly, some of the challenges that we have in healthcare.
Punit Soni
I agree, you know, and super well said, I think that few different things. First of all, maybe I’ll be wrong, you know,
Nick van Terheyden
no, no, I don’t believe it for a second.
Punit Soni
But, but, you know, a few years ago, somebody said there’s no way that we will ever have Uh, you know, anything other than taxicab setups, and transportation is never going to change. And then suddenly, in a short time, you suddenly have on demand cars. A few years ago, people said it’s not a, there’s no chance that a car can just drive itself. And now you have driverless cars coming over and running. I think that the march of technology is such that it’s slow, and then it’s overnight. And you chip away and chip away and chip away for decades. And suddenly, over a period of a few weeks, or months or a year or two, you have the same amount of progress that you have over decades. And that’s just the way tech works. There’s a tipping point to these things. I guess the point that I’m making is that we’re reaching that tipping point in voice. Okay. And voice to me is basically one part of this general area that we call artificial intelligence. Right now, if you look around large language models have captured everybody’s imagination with CHARGE GPT. And things like that truth is large language has been built for some time now. And there are much more sophisticated large language models in large companies that have not been exposed, the only difference between GPD three and in charge APD and others is that they have built a consumer product to show what’s possible. I believe that the same conversational aspects of answering questions, the ability to understand nuances around what people are saying, having customized ways of having voice signatures per person that you can use to understand who’s who this tech already exists. The only thing is that it’s not been pulled together in a constructive way. For people. The space in which this is going to show up is not going to be consumer tech, even the consumer tech will be the place where there will be a lot of attention given to it. Probably one of the reasons why you look at Alexa and other things. And they just, even though they have captured a lot of people’s imagination have not been very viable as businesses. And we can talk a little bit more about that in due time, the place where it’ll show up is a place where the users are intrinsically used to speaking in structured ways, where the ontology of what they say is unique, not small, not big, but unique. Where there are already products in place that they have been using in the past, where their workflows are such that they are much more repeatable. The same things are done again and again. And again, guess what, which is one industry where that’s true healthcare. So in my opinion, one of the foremost places where you’re going to start seeing some of this workout is going to be in healthcare. And it’s probably why are looking at Enterprise SAS look at manufacturing, look at that there are lots of other industries. Actually, the interest level in using voice solutions is higher in healthcare than almost anywhere else in enterprise settings. Because of these reasons that I mentioned, structured, repeatable workflows, people who are used to using voice in the past, and a very savvy, sophisticated user base.
Nick van Terheyden
So for those of you just joining, I’m Dr. Nick the incrementalist today I’m talking to punat. Sony, he’s the CEO of Suki we were just talking about the revolution that’s taking place in healthcare or maybe that’s the poor choice of words, given I think, you know, as you eloquently describe it, it’s slow. And then suddenly, it’s overnight. And I think we’re still in that slow section in healthcare. And, you know, specifically around voice and, you know, I’m famously quoted almost all the time still, you know, voice was two years away for the last 10 years, and it was still two years away. I think that’s not true. I think we’ve moved past that we’ve seen a number of instances. And, you know, the current trending topic that I see repeatedly is oh, it’s ambient listening. It’s it’s solving all the problems. But it’s not it’s not been repeatable. It’s not a solution that’s really sort of delivering, where are we going to see the value add? And what are the opportunities to really deliver against this promise that’s been hanging out there, but it feels like that carrot that the donkey never gets.
Punit Soni
That’s a very good point. And you’ve definitely been at the forefront of looking at these technologies and actually helping adoption of these technologies also. So there are a few people in the market who know more about this than you do. I think that ambient is actually the word itself is a little bit of disservice to voice technology. And here’s why. Tech that actually promises way more than it can do, actually ends up hindering adoption of that tech. Okay. You have to be realistic about what can be done and what can’t be done. Here’s how I look at these technologies. There is the ability to say what you have in your head. People call it dictation, there is the ability to command and control voice interfaces. That’s what you do with Alexa Siri and things like that. There is the ability to talk and have something actually be able to put it all together in in a more cohesive format. I call that monologue. There is the ability to have a discussion with somebody and have somebody summarize it, I call it dialogue, dictation command and control monologue dialogue. These are various techniques with which you can automate interactions, a better word to use for this space is automate. Not ambient. Ambient is just literally one of the things that we do, which falls somewhere between a monologue or a dialogue. The reason this matters is that when we confuse one particular feature of voice interaction with the entire future of voice interaction, we create misplaced expectations. We know that there is a bunch of companies who actually say they have Ambien solutions. Truth is, they actually do have models that allow us to summarize something. But at the end of the day, that accuracy and comprehensive that accurateness That that healthcare demands, requires them to actually add humans to the system to make sure everything is perfect and dialed in. As soon as you add humans to the system, it’s no longer a pure software product. As soon as you add humans to the system, it’s a much more highly priced product 2000 to $3,000, at which point, you do not get the adoption you need, because you’re talking to low margin health systems to generate the kind of data that will allow you to build the machine learning models to actually move boys interface forward. So this is a vicious cycle in which we’re getting caught up where people are expecting things that tech can’t do, where people are building things that actually cannot break out into a pure tech setup. And we are focusing on one feature area when we should be thinking about a whole plethora of voice interaction models. So let me propose a slightly different way of thinking about it. What if, what if you actually had an app? Or an interface that was available to you in your pocket? What if sometimes, you told it, hey, remember to just, you know, put in this particular aspect that this was a 27 year old who had this particular issue, and I just want to make sure that I plan for this? What if sometimes, you could say, hey, just insert my typical surgical consent and make this one change? What if sometimes, you could say, I’m just going to go in case presentation mode to you and tell you here are the five things I’ve noticed. And you I read you to construct a note from it. And what if sometimes, you can say, take this convergent, summarize it and put it in an editor for me. So I can actually change it according to what I think is the real thing. These are all interfaces, that we should provide access to doctors in a way that’s cognitively easy for them. And if you can do that, then you have built, in my opinion, the voice layer of the future, thinking only one of this will solve it, or one is the panacea for everything we do, I think is a recipe for low adoption, super high costs. And the same vicious cycle that we have seen and why is we’ll always be two years away from being primetime. So So yes, it will happen. But But this problem is equal parts artificial intelligence as much as product, how do you build the right product and give it to the US?
Nick van Terheyden
Yeah, I think, you know, important points buried in in that sort of description, which was, you know, very helpful for sort of understanding, especially for those that sit outside this, that over promise under delivering, I think has been an ongoing issue and challenge continues to be not exclusively to voice to be clear, but you know, certainly one that I think we’re challenged with, and sadly, it impacts is negatively downstream. Where we’re doing better at this point, we’ve got better opportunities, better technology, as you think about the rollout of this and you know, you cited an example, perhaps as a product, it’s on my phone, it sort of brings up one of the constant push backs I get great, it’s somewhere else. How is that integrate? Because in healthcare, you know, and I’ll quote an old radiologists friend of mine, if it costs me one nanosecond more, I ain’t going to use it. And if going somewhere else is that cost? It’s a problem. How do we address that?
Punit Soni
Yeah. You know, it’s interesting. Dr. Nick, I don’t know if you remember, but when I started this company about five years ago, I think The first or second year I had a conversation with you. And I still one thing that stuck in my head in that conversation was you said it repeatedly said, Remember, healthcare tech is replete with corpses of companies that basically keep promising shit. They promise stuff and like it has never get there. And you said it probably at that point as a warning, and maybe as a suggestion that if you’re going to do something, then make sure that it’s real. Because nine out of 10 times, it’s very hard to actually make things reality here, I’ll tell you why the reality is hard. It’s hard not because the AI technologies haven’t gotten to this place, or you can’t build beautiful user interfaces, it’s hard, because you have to make it work with the rest of the healthcare ecosystem, specifically, the EHRs. And if you can’t actually put the data into the right places, relatively automatically, you’re actually creating more work for people not less work, which means that even though it looks like a really sexy product, it’s not going to actually be real. And so the way I look at this is the following. I think, first of all, the plumbing of healthcare tech products is more important than the actual core AI, you should build the plumbing and make sure it works, then the AI will be obviously it will build itself because you will get the right data and be able to do things. Number two, I do not think our EHRs. And with all due respect to all the major EHRs are good enough yet to be able to embed this kind of AI technologies and techniques inside them and have people use voice interfaces and have these, these, these UIs move forward. So the key is that you provide the UI control over the UI, which means you have to have your own apps, but then you have to still put the data in a structured format and the system of record, if you can do those two things, build really beautiful, gorgeous straightforward user interfaces, put data in the right places, and make sure that you promise them only what you can really deliver from pure software, you have a shot, not an easy shot by any means. But you have a shot. And I believe that those are the things we have to keep in mind if you’re going to build something, invoice and healthcare tech.
Nick van Terheyden
Yeah, I think great points there. You know, we’ve, we’ve sort of challenged that infrastructure, you know, that electronic medical record, which, you know, I’ve also been known to hold up my hand and, you know, claim a little bit of responsibility, and not in a good way, for actually foisting this on my colleagues. And, you know, we failed miserably in the implementation of that, because what we did was we took what we already did on paper, and just turned it into technology, which was just woefully inappropriate. And of course, we’re paying for that now and trying to sort of retrofit as you think about the future, you’re obviously I mean, it’s clear, there’s clear opportunity, clear potential, not an easy path. I think we’ve got better interoperability, how do you see this all playing out? Where is the sort of pathway and think about it from, you know, the small steps that people can take, how do they get to this point, and reduce that burnout? What are the opportunities here?
Punit Soni
Yeah, no, it’s a really good question. And I will say it’s not easy. I definitely had a really Jet Black Beard when I started this company. And now five years in, you know, I’m growing like most presidents do, unfortunately, not succeeding like any of them. But the they there is a path and you know, now I have to give it put it with a pinch of salt. Everybody thinks their car is the best. So you know, obviously, you know, the way I have actually built Sookie out might look like, in my opinion, the best but there are many multiple ways to actually do this. path is to one always build at the at the cutting edge of technology. But not more than that. Path to the path is also do not actually build tech that actually uses humans and hardware. Because in healthcare, especially in current phase, it’s important impossible to scale it. And the problem is a scale problem building solutions that work for 10. Orthopaedics people somewhere is okay, but it’s a scale problem we have Part Three actually think about primary care. Make sure that primary care physicians are taken care of their 50% of our healthcare system. And if you can actually solve for primary care, you will probably find you can solve for higher end. specialties. Anyways, more importantly, as you actually build the core technology, pick a workflow end to end. So for example, if you’re going to build something that involves note creation, remember that people create notes during the visit they create after it they create half of enduring texts sometimes create half in the end, you have to think about a product that’s flexible enough that it can work for everything and it’s still a product and finally, li more important than all the AI you can build, make sure you nail the plumping compliance, HIPAA sought to getting privacy, multi tenancy putting an infrastructure in place that can actually interoperate with EHRs. And by the way, if you push the EHRs, they help. So it’s not like they don’t help. It’s just that they’re sick and tired of the number of companies who come to them who don’t have a good solution for push them. So I think it’s possible a company can many companies can be built in this space, but you just have to start from first principles of care for the clinician care for what the real prompts administrators have focused on the EHR integration, then start building the problem.
Nick van Terheyden
Are they perfect summarization? Excellent points, unfortunately, as we do each and every week, we’ve run out of time. So it just remains for me to thank you, puni, for joining me, thanks for joining me.
Punit Soni
Thank you. As always, you know, if I start talking to you, we could talk for hours. So thank you, and thanks for all the support and advice over time. Very useful.
