Scaling AI Technology In Health Care with Notable Health CEO Pranay Kapadia | Greymatter

Hi everyone, I'm Jerry Chen, general partner at Greylock. Welcome to our podcast Gray Matter, where we bring in some of today's top tech entrepreneurs and founders and business leaders to share their story from startup to scale up. Today we're talking with Pranay Kapadia, CEO and co founder of Notable Health, the startup trying to change patient doctor interactions with. We're going to discuss the go to market strategies, learning lessons for founding CEOs and the impact of AI in the healthcare industry. Pranay, happy to have you here.

Thanks for having me.

So Pranay, you and I first met in a previous life for both of us. I was an investor doing diligence on Blend Labs, a mortgage software company that you were working at the time. And we met in one of the product reviews during my early work with Blend. And then after we made the investment in Blend with Nima, who you know well, you and cup, your co founders, spun out to start Notable Health. So tell me, what was it about Blend's experience plus something else in the world inspired you to start your own company?

Absolutely. My background has been at the intersection of technology, data and compliance. I came out of the computer science degree out of Wisconsin, wanted to understand risk and money and so worked at Intuit where I had a phenomenal career working with some fantastic people, understanding compliance software and workflow software and making that ridiculously easy to use. I had a terrible mortgage experience and so wanted to really work on how could we simplify that for the average consumer. And outside of getting a mortgage, I knew nothing about the world of mortgages. Fortunately, we got connected with the Blend founding team, pre product market fit, trying to figure out how to tackle the world of mortgages. And, and that team had just some incredible views on how we could dominate the mortgage industry, despite the fact that we didn't know much how we could actually take on the world of enterprise with consumer grade software. And it just seemed like the perfect fit to actually jump in and work with them on simplifying mortgages. And we learned a lot as a team. I met my co founders there and we learned a lot about how to simplify data collection in vertical SaaS based industries. And there's often some data collection that's happening that's actually driving downstream risk. And with lending, you want to have the simplest borrower experience possible to drive the lowest risk loans and collect that data. And every interaction, every loan data that you do collect drives better value and risk downstream. And so as we were tackling that and we were building something that we were proud of my family of physicians, had this really big pain of data collection. Think about the last time you went to see a doctor. You walked in to see the doctor, you probably were greeted by somebody on the front desk, maybe three people at the front desk. They asked you to fill out all sorts of forms around insurance, around payments, around your past history. You then got roomed and then an assistant came in to take your vitals. You then spent five minutes with a doctor, likely seven and a half, which is the median that we see today. The doctor probably spent half of that time facing the screen, entering notes, putting in orders and saying I'll see you in two weeks and then left the room. And then you spend another 15 minutes with the checkout staff. Now it turns out that there's another 15 people that actually touch that chart after the doctor leaves the room before it shows up in your patient portal. And that level of inefficiency in a market that is just as compliant ridden as mortgage, that is just as complex with the multi stakeholder interaction actually got us very, very interested.

So Pranay, you're saying In a typical two and a half hour doctor visit, you spend seven minutes to the doctor, 30 minutes with people who aren't doctors, re entering all your data three times and you get a bill for $2,000 probably a month down the line or two weeks down the line and there's.

Still three back and forths between the doctor's office and the insur insurance company on what that amount should be.

What's wrong with that? Who says healthcare is broken?

So this interesting insight that we had as we were studying this space, the entire EHR market is like a $21 billion market.

What's EHR?

It's the electronic health records. It's the system of record in healthcare. And these are massive companies. They've actually done a good job of moving from paper and pencil charts to trying to digitize them to the best of their abilities given 20 year old technology. Think about the assistants in the office, think about dictation software, think about offshore transcription is actually a $60 billion market which is sadly growing faster than the EHR market is. The market to check the data that actually went into the EHR by these intermediaries is a $240 billion market. And so when we think about healthcare inefficiencies, our fundamental belief is there's plenty of teams that can go after trying to find a cure for cancer. But we wanted to utilize AI and the latest technologies to really focus on jobs. People don't want to do to eliminate the pain and inefficiencies in this market. That's all about data capture, creating a phenomenal patient experience, creating a phenomenal provider experience.

So I want to go back to the market and the pain in more detail. But before that, you're at Blend. You learn about regulated industries like financial services and how better data leads to better decisions. And in a parallel world, healthcare, again, better data, timely data can help doctors make better decisions. And then you decide to leave Blenn with their two co founders. First, tell me about your two co founders. Adam, Justin, why did you pick each other and why did they pick you? What were their attributes to form this super team? What were you looking for in your co founders? And why did three of you complement each other so well? What words about their attributes, their skills, their strengths and weaknesses? How do you guys work together? Tell me about your co founders.

We're incredibly, incredibly fortunate to get a chance chance to work together. My co founders are Justin White, whose CTO and leads our engineering efforts. He used to lead engineering efforts at Blend in the early days. And Adam Ting, who used to lead design at Blend, now leads product at notable. We've spent the last five years working with each other and sat no more than 10ft away from each other in those five years and spent more time together than with our own families. And so in getting to know one another's strengths, in getting to know how we tackle problems and really early on, I believe we're two years ahead of any other team in this space is fundamentally because we have the same vocabulary. We have a foundation of trust, we have analogies that we can draw from that actually leads to velocity on our team, unlike any that I've seen prior. And there's also a different mix of talents in what each of us brings to the table. I remember sitting around the table talking about healthcare and our ability to deal with ambiguity. Going into markets that we don't know much about with humility, wanting to learn, almost childlike curiosity, and trying to draw from partners, from customers, from advisors to identify what we should do is something that we're uniquely positioned with and we're very comfortable with as a team. In addition to that, having a support system where we're continually pushing each other to do our life's work has been incredibly valuable. Startups are really hard. We've gone through experiences where we thought the company would shut down. And to be able to go through that together and actually talk about how can we get better as a company, how can we support each other? How do we build in the system so that this never happens again? Has been instrumental. And so a good example of this was learning about enterprise integrations in a prior life and learning about security in a prior life in working with Blend where it is incredibly, incredibly complex and then being able to bring that to noteable with the exact same infrastructure and learning so that we can get a six month head start. So when we're actually in an enterprise conversation, we know this is where HIPAA stands, this is where ISO 27k as a standard. These are security standards that one must abide by to get enterprise contracts done. We just have that vocabulary. We're like, we gotta do this next month. We know exactly what it takes. We can hit the ground running to close massive contracts and not just be working with small practices. We're in this unique position where we actually get to work with practices that are like 50 physicians large, as high as 6,500 physicians large, and to be able to scale and build a product that actually could support both of those are skill sets that we've developed over time in working together. And so that gives us an unfair.

Advantage as an investor. People often ask me, what makes the ideal founding team and what kind of DNA do you look for? There's no right or wrong team. We tend to find people that are A passionate about the problem. Startups are hard B, they have some insight about the industry or the problem to solve it and then the ability to communicate and make decisions. But if I could design an ideal team, I think picking one product person, one designer and one technologist almost has all of the bases covered. Especially in a world like healthcare where the strong product data is necessary. The application of AI technology is paramount to making a change. But the design aspect, I think Adam brings around designing something that's delightful, both the doctors and frictionless and invisible to the patient is key too. So having that DNA from the beginning I think was super influential as the kind of the core stem cells of notable, if you will.

Me and my co founders find ourselves in this really fortunate position where we've had the chance to work with each other in building companies and scaling them from 10 people to over 200 in short order, seeing everything break down, but knowing that we have each other's backs, I think the trust and foundation is paramount. Having complementary skill sets was the other one that has actually worked really well for us. In, you know, my background in working with software, but also being very much product minded. How do you actually balance the different stakeholders? I Also learned a lot being part of the blend sales cycle in how do you actually sell to the enterprise? So that's been part of my role here at Noteable. And then with my two co founders really trying to fill how do we actually make sure that we build an incredible engineering culture in everything that we do from day one? And then lastly with Adam, my co founder and leads product, just really have a good taste for what it takes to drive enterprise software. Because if you just go into enterprise trying to solve for the checkboxes, you've built something that will be outdated and no better than the legacy technology that's there today. And so for us to actually build something that we're proud of, those are the skill sets that we found working really, really well together.

How do you guys handle stress or tough decisions? Give me an example in the early days of a major decision or disappointment that you guys had to navigate. And how did the three of you decide? Is it you delegate decision to one of three or do three of you guys vote or feats of strength?

I think because of the shared vocabulary and values that we do have and the common alignment on the fact that we actually want to do our life's work together leads to us having just a phenomenal trust. And so I'll give you a great example. When we first built our platform and it was a wearable driven solution, we couldn't get audio off of the watch for two hours after the encounter. And we thought it was a function of we made the wrong bet, we actually built the wrong thing and we spent three weeks of our lives doing it. Now again, that seems like a short period of time, but it was a very simple what's the right solution? Who's the right person to answer it? Defer to the person that actually is closest to the work. And with the three of us, those decisions have been really, really easy to make. There hasn't been any conflict on that, given the role clarity that we have early on.

So you've identified that the patient doctor interaction is fundamentally broken. How is Notable actually attacking this problem? What do you guys do to solve this problem? Where do you guys start?

So Notable does three simple things. What we found was physicians have devolved into being the highest paid data collectors in the world. What we want to do is actually eliminate the drudgery that is putting them behind a screen where they actually have to type things, collect information from the patient and really create an invoice because that's really what they're trying to do. We've been looking at how can we put that on autopilot? By using their own interactions and past data, we've been looking at how can we use voice as a modality? Think of this as a proactive assistant that's always helping the physician out every step of the way and learning from their patterns.

Is an app on my computer, my phone, my watch. What is notable and sounds like magic.

It'S meant to work like magic, and often the best products do. The way that we've actually built it is notable is a platform that actually sits on top of the electronic healthcare record systems. We study continually what the physician's past behavior has been in order to predict how they'll actually tend to practice. We then provide the ability for them to actually use voice as a modality to enter additional information in. And then lastly, we actually use algorithms behind the scenes to actually help them with their invoice generation for coding purposes. These are all the administrative tasks that the notes actually go into. A good analogy for this is think about your phone, where you actually wake up in the morning, or you sit in your car and it's learned your patterns. It actually says it's 10 minutes to work. You didn't have to put that in. You could tap on it and it takes you to work. Or you can just dismiss it outright. And it learns over a period of time. Tuesdays I go to San Francisco, Wednesdays I'm in San Mateo. And it actually can learn that that's exactly how physicians actually need to practice and want to practice. But unfortunately, legacy technology hasn't enabled them to do that. Now, if you want to go somewhere else actively, you could still say, hey, dingus, take me to a certain place. And your phone will do whatever it needs to map you to the right place. And so we actually support both those modalities where we found it's just as important to be proactive in how physicians tend to practice as it is to be reactive to what they're actually asking for. Here's an interesting anecdote on that. Actually, when we first started, we actually spent about six months studying interactions. We studied how physicians tend to practice. We probably sat through about 2,000 encounters just to see and follow the data. How's it actually being collected from the time a patient walks in to how does the physician actually get paid? Where is this electronic healthcare record going? In an instant, we knew that voice was going to be a modality and something we wanted to be a part of. And I remember actually sitting down with Adam and Justin and we started to build prototypes luckily, all of us code. And so we actually came up with our first prototype and it was actually an Alexa like device that sat in the office, in the encounter room, in the visit room. And I remember Adam looking at it and saying, that's too slow, that won't work. And as we actually followed how physicians tend to practice, not everything was actually being said in the visit. It might be a situation where the doctor's actually talking to you and says something along the lines of, pranay, I really need you to get 30 minutes of exercise three times a week. And I would encourage you to do something around mind wellness. But then on the way out in your chart, they'd actually put in a note that says, pranay is mildly obese. They'll never say that to you. And so Adam was like looking at this from a design perspective and his big thing, his big aha was we've got to do something that can be with the physician, that could live with the physician wherever they are. And then we tried the phone and we found that most female physicians actually don't take the phone into the clinic with them, into the visit with them. It's in a bag in their lounge. And just adding another screen would be the worst way to actually develop something. And so through iteration and some terrible ideas, we landed on a wearable and we actually sit on the physician's wrist in what we believe is the most intuitive interaction that healthcare has ever seen.

So, Pranay, after you left your previous employer to start this company with Adam and Justin, the three of you spent a lot of time thinking about the space, researching it, and then to take this plunge into a healthcare space that seems daunting. A lot of young founders out there want to make the plunge, but aren't sure what are the milestones or the roadmap markers that they should look for. When do you know it's the right thing to do and start a company right? When do you keep going back to this itch? Is it an emotional decision or is it an academic decision? How much is data driven versus gut driven?

Maybe it's a little bit of both. To think back, we were always asking ourselves, would there be a translation to what we had learned in other compliance ridden industries into healthcare? Would there be something that we could apply a formula that we had done before into healthcare that actually makes things easier to use, that uses data for better decisions and then integrates really, really well into the enterprise? And then we looked at healthcare. We found it to be not only personal, given my family of Physicians, my co founders, also have families that have a tie to healthcare administration and healthcare malpractice. And so in us just exchanging stories around that we knew there was something here. But the thing that actually made us really excited was the fact that the electronic healthcare record system had actually hit a 90 plus percent penetration. And so we weren't having to fight the first wave of how do we get people to use a computer. There was at least two years of data for every physician that we could actually tap into. And especially as you look to your data being the new electricity, how do we actually collect the data? The fact that there was certain types of data that we could tap into in mine actually made it super valuable. And then there was this moment in time where the burden of data collection had become so large. I remember the three of us getting together and saying, let's create a prototype. Let's see if we could sell it to one person that would part with money. And that's when we knew that we actually wanted to go in where the early customers actually matter a lot. And trying to figure out who would part with money for something that you can create a value versus let's do something that is intellectually stimulating actually made a big, big, big difference for us. And when we did our first encounter, I mean, we're ashamed of that product.

Well, Reid Hoffman, my partner, says he owes you to be embarrassed by the first product. Right?

We were ashamed of that product and we still have it and it still works. We kind of shipped that out to make thousands of dollars from a physician for the year without having the rest of it built out. And then when that started to spread from one to the next to the next, we knew we were onto something.

One of Reid's blitzscaling lessons is to ship your embarrassing 1.0 and then iterate, iterate, iterate, but focus on this customer success to make it better. And he'll argue in the early days, do things that don't scale, right? Which is basically you ship it your hand to hand combo with every doctor, every physician interaction until you make it better. Because at some point in time you're going to have a product that scales and you can't talk to every doctor one on one.

Yeah, there's an example of this that we ran into. We started to go live with physicians and initially we actually did. We just copy pasted data into their EHR because we didn't have any integration. The road to limbo in healthcare is driven by a whole number of companies that haven't been able to integrate into these large monolithic applications. And we thought we could work with an integrator and that would just work. We were copy pasting. It wasn't scaling. We're now doing hundreds and hundreds a day, and we just couldn't do that. And we're like, we need to integrate so that this can go away. And we found an integrator. And I remember we sat down in the first meeting with it for a group of 50 physicians. This is not a massive group to integrate into their legacy systems. And the roadmap that we got was it'll take nine months to be integrated into their system. In my head, looking back now, I think Reid has this really good philosophy of scale over efficiency. And we're just like, how do we get scale? How do you create a curve in healthcare that isn't ridden by long IT roadmaps? And we studied the systems and we learned. You taught us this. On enterprise integration, one of the key things that actually stood out was what we did with Mint and TurboTax at Intuit. We integrated with 6,800 banks using robotic process automation. And we can now go live in less than a week fully integrated in any of these legacy systems. And the beautiful part is we've actually found that that scale over efficiency is actually scale and scale.

I learned that lesson from my partner, Neil Bushry, who said, you have to show customer value quickly. And in this day and age, especially the consumerization enterprise software, we're used to instant gratification. No customer's waiting nine months to see value to the fact that you do what you need to do to get value in like hours, weeks, days, and then you can scale later. But you realize that scale and speed is actually a skill.

And you know, it's interesting in healthcare is the CIOs. I was actually talking with one recently in Atlanta that we're going live with, and he's like, well, mostly my job has been Cino, and I just have to say no to absolutely everything. But I just want you to know that it's not in my best interest to do what physicians want or need in probing on that. He's like, well, I just have to worry about the bottom line. I have to worry about risk of data. But when we actually talk to him about, okay, well we're just going to use, give us accounts on your system and we're off and running. He's like, wait, you don't have an IT project. There's no integration here. And that's actually been really, really helpful for even the CIOs to actually tell their physicians we're bringing technology to life so much faster. So it's not just like consumerization of the Enterprise, it's also CIOs that are starting to see I want the value right away. And I would look like a hero if I do that. Exactly.

One of my adages in enterprise sales is one, don't get the customer fired and two, make the customer look like a hero. So I think what notable does, it makes the customer, in this case cio, the healthcare provider, look like a hero. So we've looked at a bunch of these smart agents, the Siris and the Googles of the world try to do this. They do, I guess, three things. One, they try to prep you, say, hey, appointment coming up, you're 20 minutes to this meeting. So from what I understand what Notable does, it actually reads the EHR data, patient records and kind of pre populates prepare the doctor correct with a kind of a virtual chart if you will, before the patient interaction, then during the interaction. The way I understand it, Notable actually lets you use voice as an input mechanism. So doctors can actually make commands or instructions to follow ups and then after the fact, which is, I think one of the powers of Notable is behind the scenes is doing a lot of the drudgery of billing and data entry. And really these early EHR systems were billing systems.

Right.

They were not ever designed for patient doctor interactions. They're meant for patient doctor insurance billing interactions.

Actually they don't include the patient at all. If you actually ask, 9 out of 10 doctors will tell you that the reason that they chart or they do their notes, number one is malpractice to cover themselves. Number two is to get paid. And number three is for continued care of our patients. Nine out of ten doctors will tell you that in that order. And it's not because they're bad individuals, it's that the technology hasn't assisted them to actually focus on what they want to do. So that's the part that we believe has to change with easier to use software, easier to use technology that collects better data to drive better decisions downstream.

So we always talk about this unintended consequences. We built these software systems to automate billing for doctors. And what happened is these systems grow up to now run how doctors interact with patients. How do you avoid the trap of notable? Not just making billing systems more efficient and speedier, how do you not make our robot overlords better and faster, but how do you actually break the system? And my favorite analogy is Google has taught a generation of computer users not to ask normal questions. We now speak and type in keywords. Right? And so these systems have now taught doctors how not to interact with patients as human beings, but rather billing codes. Right. Broken leg or pain or headache or disease. How do you not just make the current system faster? How do you actually change the system?

We have a fundamental principle in everything that we build, which is anything we build, we would like to be the patient on the other end. And that's how we tend to bring design into what we do, which is if the physician is actually spending time with a computer screen trying to check off boxes, we're going to take that away, automate that, extract information from conversations that they're having. And so that's actually a big part of our principle. The way that we see the system evolving over time is step one is to actually take away the keyboard from the physician. And you do that in a proactive assistant way. You do that in an active assistant way using voice. Step two is actually to utilize the new data sources that we're actually bringing to market. We're doing thousands and thousands of encounters on a weekly basis where we actually have, in an anonymized way, what the patient and physicians actually talked about. What are the topics that they covered? Did they talk about falling down? Did they talk about smoking? Did they talk about X rays in any which way? Do they talk about opioids? And actually bringing that to life for physicians has been pretty illuminating. We've actually done some interesting insights with physicians as an example, where we give them their own heart rate at the end of the week, where they can actually see their own heart rate with types of patients that they've seen, and that's actually driven better behavior, where they actually know they want to go revisit certain patients that actually elevated their own heart rate. Now, we also found that the heart rate has some relation to billing levels, but those are new data sources that we found that are new and game changing. And then the third part of it is, we believe, and we'd like to see the industry move in this way, and it'll take years to do, but we're already starting to see some of our customers doing it. If you've ever looked at your own health record, you know, mine was unremarkable in three places. In fact, the word unremarkable is one of the most commonly used terms in health care records. And that's where notable really comes from, which is we want to change the unremarkable into remarkable. But the words and verbiage that's actually put in those health records are really difficult for the average consumer to understand. But what they do understand is what your physician is asked to do with you, which is talk to you in seventh grade language. And so imagine a world where you see your physician, the documentation is happening behind the scenes. The payers are actually getting data in real time without having to go through three clearing houses in order to get data. And the patient, when they leave, they get a snippet of the audio of what the physician actually told them to do. So you have a better interaction with that patient or with that physician and you have something that's actually there for recall. One of our physicians is actually doing this today where he shares the audio recording from every encounter back with his patients. It's a spinal surgeon. And the net promoter score of those patients is off the chart. Because anytime somebody comes to you and says you need to get spinal surgery, you stop listening. Your life is flashing by and now you actually want to be relying on caregivers. You might want a second opinion, but by having that exact specific, what was actually discussed back home with you is something that we believe will actually change the system and not just create a faster robot overlord.

Walk me through typical patient doctor interaction and how they're interacting with notable from the beginning to the end.

Sure. So the way that our product works is imagine you walk in as a patient into your doctor's office. Notable actually studies what's bringing you in and how the physician that you're about to see tends to practice for similar looking patients. We use that data to pre fill your chart for that physician such that they have less work to do in doing the mundane tasks that they would have to do. When you do go in to see the physician and Noteable actually sits on the physician's wrist as a wearable and is listening into the conversation in a very ambient way, extracting topics and insights from that conversation. What were the drugs that you guys discussed, what were the ailments that you may have discussed? And extracting that information to identify what work we can do behind the scenes for the physician. When the physician actually leaves, they can speak into their watch to place any order. So it could be a prescription, it could be an X ray, it could be referral to another doctor. And we've actually created our own intent parsing system to identify. Think of it as when you say, hey Siri, do something, but you can actually talk to Notable and actually place those orders and we take care of all of that and then noteable takes all of that data, structures it, puts it in the legacy system of record and then pre fills information around the billing that's actually required for the doctor to get paid. That drives value for that group.

So it sounds amazing for the doctor to save time and preparation and automates a lot of interactions and it sounds like the patient can also get better care because he or she can access the data, the recordings and have a testimony on hand what happened. How do you think about privacy? How do you think about this whole regulated industry? It seems like it kind of intrusion between this almost sacred interaction between a patient and a doctor.

Yeah. And these were actually lessons that we learned in working with FinTech and working with heavily regulated industries is that the data is actually the patients and the providers. It's not notables. We do anonymize the data that we do collect in a snapshot of time such that we can't actually go back and identify which patient this refers to or which physician this refers to. And we've actually spent a lot of time in how we can anonymize that data because at the end of the day it's that training data that actually helps create this autopilot like environment and makes that possible. So as a company we very early on invested a lot of time in setting up the security infrastructure, setting up the anonymization frameworks that we currently have around this data and setting up the integration infrastructure.

Autopilot for doctors. Sounds like a great way to describe it. Maybe copilot is probably the better way to describe it so doctors don't feel too threatened about their jobs. Let's talk a little bit about what happens next. Because you build this product that kind of automates their interactions and there's clearly three parties involved. There's the physician, there's the patient, and there's the payers, the insurance companies, which are kind of the thousand ton gorilla sitting over the entire healthcare industry. When you actually build a product, two areas I wanted to talk about. One is how do you quantify value? How are you selling this? How do you quantify value? And number two, when you think about selling it, who are you selling it to? Are you selling it to the doctor, individ, the hospital, the insurance company, the patient, all above. So first, how do you quantify the value and who are you selling this to?

The unit of value we look at is really how many encounters are we powering and to do thousands and thousands of those on a weekly basis is what gets us to collect more data. And then using that data, we look at how many touch points does a doctor actually need to have internally we talk about what does the one click encounter look like where all the physician actually needs to do is sign a note. And so think about this as the analogy being self driving cars, where the data that you're collecting is how many touches did you have on the steering wheel for every thousand miles. And so that's kind of like the analogy that we use internally. We're trying to make sure that we're automating things in a way that drives value from a sale perspective. We're building an enterprise first company. That is the goal. There's more and more aggregation happening in the healthcare industry as a whole. And so we really sell to provider groups and provider networks. The efficiencies that we bring them, the time savings that they have and the potential added income levels that they do get has been really helpful. As we see this evolving, we're also seeing a trend where a lot of aggregation stems beyond just the provider groups and payers are starting to actually purchase. So think about insurance companies buying practices and owning the doctors, employing the doctors. And so as we see as that trend continues, we believe we have even more value to add because of the data from every encounter, which is so much higher fidelity than what's actually in the EHR today.

So if I understand what you're saying, the ultimate thousand interactions autopilot. So if I am self driving car, I am justifying the cost by lives saved and productivity. I can check my iPhone and text message instead of driving because I never check my phone while driving. That would be unsafe for notable. The value then is I'm saving doctors hours of time. I think you alluded to two hours. For every one hour of patient time, you're going to save them two hours or three hours each day. So that's increased productivity for the doctor, but also probably just decreased burnout, less paperwork. Yeah.

So there's this emerging industry and it's kind of a wild concept in our minds of scribes and we actually don't like that word.

What's a scribe?

A scribe is a human assistant of sorts that actually follows the doctor in every single interaction and types on their behalf. So it's a band aid solution to get the physician out of the keyboard and the screen. What's interesting about that is if you actually Google that term, it's an 18th century job description whose sole purpose is to create copies of the Bible before the printing press. So we don't know how to do that we don't want to do that. That is not a job that AI needs to solve in making copies of things. And I was actually with a large health network in Ohio recently and their CEO made it very clear everyone here can get a scribe because physician wellness is a big problem. But in order to get the scribe, you have to see four more patients every day to justify the cost. And the big aha there was we're adding a band aid to an industry and then asking physicians to run faster as hamsters, which is actually not the solution. One interesting aha for us was on average four and a half support staff per physician today. And each of those supporting staff, and this is the front desk, the billing, the coding, which are between 50 to 75,000 ahead per physician today in the healthcare industry. That's what we want to go after by collecting better data, not asking the physician to run faster. Yes, some of our physicians actually do see more patients. They take home $2,000 a week more, which is a fantastic ROI for them, but that's their choice. There are also physicians that say, I'm done at five, I don't want to see any more patients. I want to go blueberry picking with my daughter. And that just is a profound impact that we can make where it's not about asking doctors to run faster and asking physicians to work harder just so that we can make your life easier with technology.

So we've created this system where we have software to make you more efficient, but to use the software we hired human scribes to pay for the human scribes to force the doctors to work more hours. Yes, this is loop, this vicious cycle where doctors are working harder so they can work less, which is the irony of it all. But I think goes to the crux that you're attacking not the 21 billion of EHR spend. You're not just attacking the $60 billion of data and billing spend in healthcare, you're attacking this $240 billion of cost to input, check and clean the data from beginning to end. And if you can take a fraction of that money out of the system, not only do you reduce health care for our entire country, but in turn make doctors more productive and kind of enrich this patient.

Doctor interaction, we spend more on healthcare billing, administration because of the inefficiencies of collecting data upfront than we do on education and roads in this country.

So let's talk more about the different ways notable is enriching this interaction. You mentioned there are some gleaning and learning, some AI applied to collecting the data off the HR system. And voice or natural language processing, NLP is another tool too. AI is kind of this buzzword we hear in almost every single pitch these days, but you're actually using legitimate technology to both interpret information and voice to text as a data input. Talk to me about how do you think AI is going to change healthcare? Not just a notable, but writ large. What trends are you seeing being the front lines?

If you think there's a skepticism in the valley around AI, you should talk to doctors around AI therapeutics in any which way. And it's really hard to tell a physician that a computer might be better than you. And as soon as you do that, they don't want to adopt it. From our perspective, we look at it as how do we tell the physician what their own patterns are? And if we get them to actually practice consistently over time, they will get better because they actually want to. They just never had access to the data to be able to do it. The second part of it is we look at it as how can we use AI to take away jobs people don't want to do first? And if we do that, we then have a right to tell people how to do their jobs using AI. And it's just such low hanging fruit in this environment that we would be doing a disservice to the entire industry if we didn't tackle that first.

We have a framework. I heard from a founder once, Terminator vs Ironman, right? Is the AI a Terminator here to replace us? Or are using AI as a suit of armor to augment the humans? And so using AI to augment what doctors do day in, day out is kind of what they want. They want the Jarvis, the suit of armor that makes doctors super productive and superheroes, rather than having these robots that are soulless kind of take over from their jobs.

And they have a general distrust for black box systems. They're very data oriented. And so by actually making them love the software, by giving them augmented abilities is actually how we see this evolving.

Well, the best technology out there lets humans do what humans are great at and lets systems do what systems are good at. So systems are really great at automating simple tasks, making really clear cut decisions or interpreting data. And humans are great for two things, the unexpected, the decision making around imperfect data and actually a bunch of interactions that you actually need flesh and blood and a pulse to interact with. And so the systems that I think use AI correctly let people play to their comparative strengths, let humans do the decision making, the cognitive Ability we need to do and let computers do all the rote stuff that computers are going to excel at better than humans.

There's nothing better than working with a cancer doctor where he can actually hold his patient's hands and look them in the eye and tell them, I know you probably won't make sense of this right now, but you have such and such a cancer. It's not life threatening. Here's what we can do about it. Here's how you actually come out the other end. I don't believe software will actually solve for that. And we don't need to. But to allow that physician to do that and have that empathy, where all the work for him or her has been done behind the scenes is where that conversation was captured. The orders are placed in for the chemotherapy. Those are the things that are actually taking away the empathy that's preventing us from us, from us being humans. And so that's how we believe and we are committed to bringing technology into this realm. And it could be AI, it could be something that we haven't even thought of yet. But we don't want to take the human connection away for where it should be there. But for all of the things that people don't want to be doing. It's kind of like how we look at that changing.

Give me an example of how notable has changed how a doctor has interacted with a patient for the better.

There's a number of them. We have this one that stands out to me where we studied how a practice worked and their patterns. And they tend to try to follow really good care pathways, which is if a patient has a certain kind of ailment, they tend to have a certain protocol of how they do that. Unfortunately, there's never been technology that can learn their history. And so we studied this and I got a call from gynecologist and she said, hey, I love that you can study my past and you predict what I'm likely going to do with the patient. You put in this certain order, which really made for an uncomfortable conversation. Now, it happened to be that this order was for an STD test because we looked at the pair, we looked at the age, we looked at why the patient was in and clustered her patients. And this is how she tended to practice. And so we pre populated her chart with certain things. And so she calls me and she's like, this is awesome. But I know this patient, I know their family, I know their kids. It wasn't right. And I was like, okay, well let's just delete it. It's one click to delete, 16 clicks to add. And she's like, okay, got it. It saves me an hour and a half a day. So you're telling me if I delete it enough times, the system will learn? And so we went through the day. She called me a second time, and no physician will call you twice in the same day. They are unfortunately paid on time. And so she called me a second time in the day, and I was like, hey, I'm really, really sorry. I'm not really sure what we did. And she's like, hey, that STD check that you added, keep doing it. She was cheating on her husband. That's how I want to practice. That's who I am. Just the fact that I knew the patient actually led to me having certain biases. But I know that this is my data. This is how I practice. You're not telling me how I should be practicing, and therefore I can trust it more. And so we actually really see that being profoundly different in that you can win the trust of the user and then build a better system for everyone.

That's amazing that Notable is able to predict or recommend the check for STDs. The doctor probably would have skipped, and that would have been the detriment to the patient. And what Notable did is it said, hey, actually, you should add the STD test, because given the history, given the profile, given what's going on, and given how you practice, this is what you should do. Don't let your own biases interfere with what's the right thing for the patient.

And this just circles back to enriching every single interaction that they have with data from every other interaction that has actually never been done in healthcare before.

So I remember when you, Justin Adam, came out of Blend and started talking about this idea, and you were looking to raise seed money. And I think we connected because my colleague Sahm and I wrote this blog about a system of intelligence and system of engagement. And what we saw in Notable was two things. One, you recognize a system of record was the ehr, and that wasn't going anywhere. But number two, you were looking for a new system. Engagement Voice being the first of a better way to interact with the EHRs when you felt like this was the right wedge voice. We're being trained from Siri and Alexa as a way to engage with systems. But on top of that, you took that system of engagement voice and layered on what we call the system of intelligence, the ability to read the data, predict what the doctor needs to do or say. To the patient, get inputs from the system engagement and then output a better, more efficient product for the doctors that really fit this virtual cycle around system of intelligence that we were looking for, combined with the system engagement with voice that we thought was going to build a very powerful moat. And so I think over many whiteboard interactions with Yumi, Som, Justin and Adam, we finally got to this vision of what notable 0.1.1 would look like. And we were really excited to lead the seed round from the very beginning. And I think what you guys have done over the past year and a half has been nothing short of stellar in terms of thousands of weekly engagements, doctors from dozens to thousands, all basing their practice now upon notable health. So let's talk about the stage where Noble is right now. Greylock led the seed round that got you from this inception through this early MVP and early customer traction. So you recently raised a series A from some great investors in the healthcare space. Carl from F Prime and Annie Lamont from Oak. What are you going to do with this influx of capital? What does the next lap or next act for notable hold for the future?

We're in this fortunate position where we have incredible investors that teach us not just about enterprise software, about blitzscaling that has been incredibly valuable, but have also been some of the leading healthcare technologists, healthcare investors. And so we're able to tap into this massive network and learn from them vicariously based on some of their sins. That's actually been helpful. The place where we find ourselves today is incredible amount of demand, some amazing partners, and when I say partners in the true sense that they actually want to work with us and this will be game changing going into next year that know that we're in our infancy and are supportive as we grow out the technology, but we also find ourselves in this position where we actually don't need a sales team right now. And so we want to actually take this capital to increase our velocity, our hiring on engineering so we can go after more specialties, we can go after more partners and really power every single physician patient interaction and customer success, or clinical success as we call it in the world of enterprise, in the world of workflow, we want to make sure that every physician is a raving fan that loves us and we can power 100% of their interactions. And so in order to do that, physicians are creatures of inertia. They learn behaviors, some good, mostly bad, when it comes to using software that is hard to change. And so we want to make sure that we actually have a solid clinical success. Team to help them with that. And so that's really where we're investing our time is building better product and tighter feedback loop from the customers to make sure that they're delighted.

You just mentioned around sales and right now I don't need a great sales team, but eventually you do. Let's talk about this founder led sales. A good friend of mine, Pete Kazanji has this whole framework of founder led sales. So a lot of founders out there find themselves not being sales executives but having to learn to be a sales execration early on. What advice would you give these young founders or even experienced founders around founder led sales?

It's the fastest way to get a read on the market, a read on the actual value of the product. It's something I actually really enjoy doing because I get a pulse of what's important to the other person. Where are they actually looking to improve? How should I be thinking about the roadmap internally? But it also is especially early on. We evaluated hiring somebody or hiring a couple of folks on the sales side. But by me leading the charge on that, I'm able to actually throttle knowing our internal velocity, where we should actually push and where we should pull back without any misaligned incentives across teams. It's the only way to have this amazing pulse on the customer and know that you can actually build a long lasting company.

How do you learn the mechanics of sales? Some people say sales reps or barn, I actually believe you can learn the craft of sales. But there's mechanics around tracking pipeline tracking prospects, learning when to ask for the check and negotiating that deal. These are the mechanics that technical founders, product founders, design founders have not done before. How do you learn on the job and how do you get back better at this?

Yeah, I would say I got some of the best coaching and training on that actually at Blend. So that was absolutely incredible. What we want to make sure we're doing is collecting data on all the interactions. And so every single conversation that we have with candidates goes into a pipeline. We're actually collecting data such that we can then come back and review it because at the end of the day, just like engineering, these are systems and as long as you're actually tracking the right things, you can make course corrections where even if you lose a large account, and so far we haven't lost to anyone in sales, but even if you lose a large account, you can incorporate that learning into the next one. Because the market is 3.2 trillion in our case and there's no shortage of opportunities.

I think that Attitude of the learning mentality from every interaction is key to any success. Founder and really when I look at the best founders out there, the ones I want to back is what they're called the learn it alls right. The folks that every failure, every interaction be it product, sales, engineering, investment facing, they learn from every interaction to improve. Pranay, in closing, what's one piece of advice, one framework that you would suggest your listeners they adopt they can actually move the needle and make their company better.

The way that we've talked about this internally as a team has always been we are diving into the ambiguity. There's a lot to figure out and as we think about embracing that ambiguity and actually building something of value, our fundamental belief internally is we will screw a lot up and we constantly push ourselves on don't apologize, be better and learn from every one of those screw ups because the number of times that we do that and most of them have always been my fault at the company we've still tended to actually succeed despite ourselves.

Don't apologize, be better. I love that mantra you should put on T shirts and mugs. Don't apologize, be better. I wish that I can apply that to my life every day. Pranay thanks for the time. Today on Gray Matter and Alt Insights it was amazing to learn and talk about thinking about healthcare as an opportunity, thinking about voice and and AI as a technology wedge and more importantly how to just be better in every interaction and keep learning from zero to startup to scale up as an investor. It's been great to be with you from the beginning as a seed investor and series A investor and I think that a lot of listeners out there will apply these learnings not just in healthcare but in any vertical SaaS problem out there and I hope to see them succeed.

Thank you Jerry. We are proud to be doing what we are as well as proud to be part of the Greylock portfolio.