AI transcript
0:00:02 Hi, and welcome to the A16Z podcast.
0:00:06 I’m Hannah, and this episode is all about building a software company in healthcare.
0:00:11 In this conversation, Jorge Conde, A16Z general partner in bio and healthcare, previous founder
0:00:16 of the genomics company Nome, and Julie Yu, partner on the deal team for the Bio Fund,
0:00:21 and previous founder of the patient provider matching system Kyrus, explained what it is
0:00:25 that makes building a company in the healthcare space so fundamentally different from in other
0:00:26 sectors.
0:00:28 And why exactly it’s so damn hard.
0:00:32 So let’s start with basically just the very fundamental difference between building a
0:00:36 software company full stop and building a software company in the healthcare space.
0:00:39 What are the most foundational, crucial differences?
0:00:45 Well, historically at least, software had two very important sort of qualities in healthcare.
0:00:48 The first one, the actual quality of software deployed in healthcare system historically
0:00:49 has not been great.
0:00:51 User interface wise and experience wise.
0:00:52 Bad track record.
0:00:53 Bad track record there.
0:00:56 And the second one is that it was usually not highly valued.
0:01:01 So at least a lot of times it was considered either free or cheap.
0:01:02 And why was that?
0:01:05 That in it from the very beginning, there was not a lot of value attached to this.
0:01:09 On the healthcare system, a lot of things still have a very human component to them,
0:01:14 automating things and sort of creating frictionless experiences or delightful experiences.
0:01:17 The things that software is really good at doing, it’s just really hard to do in the
0:01:18 healthcare system.
0:01:21 The second one is, I’m going to generalize for a second, but I think a lot of times in
0:01:26 the healthcare system, software has sold us a component of a broader service or of a
0:01:27 broader offering.
0:01:31 And so therefore it’s the piece that tends to get sort of devalued first because it obviously
0:01:32 has the lowest marginal cost.
0:01:36 It’s going to create this weird dynamic for software companies that are trying to build
0:01:37 in healthcare.
0:01:41 There’s a higher degree of sensitivity in this particular market for things that get
0:01:44 in the way of the patient provider experience.
0:01:47 One of the challenges/opportunities within healthcare is that it tends to be much more
0:01:50 risk averse when it comes to adoption of new technologies.
0:01:55 One meaningful difference in introducing a software product to this market versus other
0:02:02 markets is the level of scrutiny and the bar that you need to hit from a not even usability
0:02:07 perspective but just utility and actually having validation of if you are going to introduce
0:02:13 something new into the care delivery flow, it better work because the stakes are so high.
0:02:16 If you get it wrong, you could either send a patient in the wrong direction or they might
0:02:20 not get the care that they need or it could actually harm the individuals involved.
0:02:23 So not just higher barrier to entry but higher stakes, correct?
0:02:24 Immediately.
0:02:26 They’re a reticent buyer, generally speaking.
0:02:30 They’re running on very thin margins if we’re selling into the healthcare system, into provider
0:02:37 space and it needs to work because if it doesn’t, obviously there can be patient harm so the
0:02:42 probability that a newcomer, an upstart can come in and make that case in a convincing
0:02:44 way is a very, very difficult challenge.
0:02:48 So does that mean you have to have certain prerequisites that you may not need to have
0:02:49 in other spaces?
0:02:52 If you know you have these challenges and you know that you’re entering this space with
0:02:55 a lot more barrier to entry and a lot higher stakes, are there certain things you need
0:03:01 in place, a certain kind of proof of concept that you might not have to have otherwise?
0:03:04 Well, first of all, I think you’re touching on a very important thing which is in the space
0:03:07 and I’m going to specifically focus on sort of the healthcare system.
0:03:11 So let’s call it provider systems, payers and the like.
0:03:17 You have to really understand what the workflows are, what the problem space is and how to
0:03:19 actually address any of those things.
0:03:24 And so one of the biggest challenges I think that companies have when they want to build
0:03:28 software products here is to really understand what problem they’re going to solve because
0:03:33 I think you have this weird sort of intersection between it’s very non-intuitive, it’s still
0:03:39 very human driven and centric, there are regulatory barriers, you don’t want to get in between
0:03:42 say a provider and a patient, you know, most people aren’t born with the ability to say
0:03:48 like I know I can insert a piece of software into this part of the workflow and I will
0:03:50 solve an acute pain point for the system.
0:03:51 That’s not obvious.
0:03:53 And some of that is actually lack of standardization.
0:03:57 You would think that medicine is an industry that has a tremendous amount of standardization
0:04:01 and protocols around how people make decisions and do things.
0:04:04 But it actually turns out that healthcare is an industry that actually is characterized
0:04:06 by a tremendous amount of variation.
0:04:07 And variation in what kinds of ways?
0:04:11 It could be variation in terms of actually literally the decision that if you have ten
0:04:16 doctors who are all presented with the same patient, you might see ten different decisions
0:04:18 about how to treat that patient.
0:04:22 Some physicians might be more aggressive about using invasive surgical techniques versus
0:04:27 others who are more holistic, even just how I was brought up religiously or culturally
0:04:29 might impact the way I think about that problem.
0:04:33 From a product perspective, you could have multiple drugs that all treat the same condition,
0:04:35 that all have different implications and whatnot.
0:04:39 So even there, even though you have a patient population that is characterized by the same
0:04:43 diagnosis, you could have dozens of different ways that those patients play out.
0:04:47 And so it makes it very hard for a technology company to come in and sort of generalize
0:04:53 and say, you know, there is one single method for, you know, manufacturing this thing or
0:04:57 for making this decision and managing this patient population, ultimately that reflects
0:05:01 as differences in the financial profile of different patients.
0:05:04 Healthcare, it’s like politics, it’s very local.
0:05:09 Thinking that you’re going to have an out-of-the-box, one and done solution, even in systems that
0:05:14 look similar from either a size standpoint or reach standpoint or even a geographic standpoint,
0:05:16 these are all kind of end-of-ones.
0:05:18 So what does that mean?
0:05:24 So we have kind of knowledge of workflow, the knowledge of variety and spectrum and that
0:05:27 you are ultimately working in weirdly an N=1 scenario.
0:05:31 I want to bring it back to like actual practicalities of this sort of company building.
0:05:38 In your experiences, you both founded companies, what do you wish you had known or done differently
0:05:42 from the very beginning, given the complexity of that space and the unique challenges that
0:05:44 building a company in healthcare presents?
0:05:48 With Kairis, one of the products that we had was a product that was used by call center
0:05:50 agents in hospitals.
0:05:53 And our thesis when we first launched the product was, oh, well, we’re just going to
0:05:58 go after every hospital that has a call center and they probably all operate similarly.
0:06:02 And what constitutes the job of a call center agent is probably relatively homogenous.
0:06:06 And so we can make all sorts of assumptions about how it’s built, how it’s deployed and
0:06:07 how it’s managed over time.
0:06:11 The thing that strikes me already is that feels like a reasonable assessment of the
0:06:12 lay of the land.
0:06:13 Yeah.
0:06:15 And especially, I think it’s very easy to get fooled in healthcare by looking at other
0:06:20 industries and seeing how it works in the rest of the world because certainly…
0:06:21 And then you pull up the…
0:06:22 Yeah.
0:06:24 And then you pull up the wool and it’s like, oh, it’s completely the opposite.
0:06:25 Call centers.
0:06:28 I mean, that’s definitely an industry that if you look at retail or even all the airline
0:06:33 companies and how they operate their customer service operations tend to be pretty standardized
0:06:36 and pretty sophisticated in a lot of cases.
0:06:39 When did you start to realize this wasn’t maybe your average call center?
0:06:45 Like on day one, first of all, there’s heterogeneity in the actual scope of services of pretty
0:06:47 much every call center that we encountered.
0:06:51 Some call centers might be fully centralized and they’re like a central 800 number that
0:06:56 receives every call that comes into the hospital versus others that are decentralized that
0:07:01 only serve the primary care line versus the cardiology line versus the dermatology line.
0:07:05 And because of that, they will have just fundamentally different starting points of where they have
0:07:07 to be in the workflow for the thing to work.
0:07:12 The other aspect is the scope of functions that the call center plays.
0:07:15 It could be everything from just a general marketing service where a customer might call
0:07:18 in and say, do you provide these kinds of services?
0:07:20 Can you give me directions to the clinic?
0:07:23 All the way to I need a prescription refill.
0:07:24 I’ve been diagnosed with this thing.
0:07:27 I need to figure out what kind of surgery I need.
0:07:29 So again, much bigger range of possibilities.
0:07:30 Correct.
0:07:31 Yeah.
0:07:36 Like I’m a call center agent and how do you define in my job so that when I give you another
0:07:41 piece of software to use to do that job, it’s going to be seamless.
0:07:44 And when you have that kind of heterogeneity around even the sheer definition of what the
0:07:48 job is, it makes it very hard to design a scalable solution that can kind of fit into
0:07:50 all those different environments.
0:07:55 So day one, we actually were fortunate to get a customer that did have a pretty robust
0:08:00 centralized call center group that was hundreds of people who literally were answering every
0:08:02 call that was coming into the health system.
0:08:06 And so the immediate sort of leap that we made was, oh, they must all look like this.
0:08:11 Even if 80% of it was the same and there was 20% sort of buffer that needed to be modified,
0:08:12 we can deal with that.
0:08:16 Yes, they all had central call centers, but the fundamental scope of jobs that they were
0:08:19 doing were completely different across the board.
0:08:22 And some were more clinical in nature, some were more marketing in nature, some were more
0:08:24 financial in nature, et cetera.
0:08:26 So what were the knock on effects of that?
0:08:27 Yeah.
0:08:32 The impact on like, go to market, product design and spend product strategy.
0:08:36 Most importantly, the service model of you could either say, we’re going to design our
0:08:40 software to be so flexible that it could work in any environment.
0:08:46 Or you could say, we’re going to provide services to come train your people to behave in a more
0:08:49 standardized way relative to the rest of our book of business.
0:08:52 And so we ultimately ended up taking a hybrid approach to both.
0:08:55 But the latter, you know, that services approach is something that we hadn’t thought about
0:09:00 that allowed us to sort of abstract out the variation to some degree, but also provide
0:09:04 value back to the customers in a pretty unique way because then we had the best practices
0:09:07 for, you know, how it should work.
0:09:10 So ultimately it was a good thing, but it was a major fork in the road.
0:09:11 Absolutely.
0:09:15 Because there is so much variability, because there’s so much localization, the notion of
0:09:20 the pure SaaS model where you’re just throwing technology over the fence and assuming that
0:09:25 it will fit into whatever environment you’re deploying it into, that is a moot point in
0:09:30 healthcare, you actually do need to think about the services component of things.
0:09:34 There was a whole generation of companies that got started like a decade ago that took
0:09:39 these sort of tech-only approaches and failed to get scale or had to fundamentally pivot
0:09:43 their models to actually take into account more of the human element of the service delivery
0:09:44 model.
0:09:45 I mean, even there’s a term for it now, right?
0:09:50 Tech-enabled services is a way of doing things now in digital health that I think is well
0:09:56 recognized that it’s necessary to wrap the technology with a human component to essentially
0:10:00 address and be able to accommodate all the variation that you see across different customer
0:10:01 bases.
0:10:04 And it changes your cost structure fundamentally, the nature of how we talked about the business
0:10:05 and how it scales.
0:10:08 And even our fundraising strategy fundamentally changed because of that.
0:10:13 And so we did have to, you know, raise more and give ourselves more runway and think about
0:10:15 different ways to manage our margin.
0:10:19 It sounds like everything that could have been changed by that.
0:10:23 Let’s go back to a specific example where you really put your foot in it.
0:10:26 Well, so in our experience at Noam, it was interesting because here, this is a company
0:10:32 with, the sole purpose of the company was to provide software capability to analyze
0:10:33 genomic information.
0:10:36 And so, you know, when you launch that, your assumption is, well, this could be used to
0:10:39 power all kinds of applications.
0:10:43 It could be used for research, either an academia and industry, it can be used for, you know,
0:10:44 clinical diagnostics.
0:10:45 Flexible.
0:10:46 We thought it was very flexible.
0:10:50 But challenge one is, you know, a solution looking for a problem is always a very, very
0:10:51 dangerous thing.
0:10:52 I think that’s universally true.
0:10:54 I think it’s especially true in the healthcare space.
0:11:00 And challenge two was understanding exactly where, in the case of the clinical setting,
0:11:03 where this technology would be used in the workflow.
0:11:07 So here we wanted to go after the clinical labs.
0:11:08 That was your initial hypothesis?
0:11:12 Our initial hypothesis for an application in a clinical setting.
0:11:17 You have technicians and docs that are inside of the laboratory setting, receiving samples,
0:11:23 running a test, analyzing the results of that test, generating a report that gets signed
0:11:25 off by a lab director that goes back to a physician.
0:11:27 Usually it’s in the form of a diagnosis, right?
0:11:29 And it gets signed off and it goes to the physician.
0:11:36 The physician now takes that report and basically decides what to do based on that information.
0:11:42 So our assumption was, well, if you have the ability to sequence DNA now in a way that
0:11:46 you couldn’t before, before you’d have to do all of these specific tests, you have to
0:11:49 know what to test and then you’d test it and then you’d get a report.
0:11:53 You had to know what street lamp the keys were under, right, like there in that case.
0:11:57 Whereas once you had the full genome, you could just sequence everything and just run
0:11:59 a bunch of software queries.
0:12:04 So our thought going into this was, well, that’s an incredibly powerful tool for clinical labs
0:12:08 because first of all, you can sequence just once and analyze over time.
0:12:09 Right.
0:12:12 It seems like a totally legitimate assumption to make.
0:12:15 And it turns out that there was a lot of challenges with that assumption.
0:12:17 The first one is every lab is different.
0:12:21 A lot of them didn’t have the budget or the willingness to basically pay the upfront
0:12:28 piece to buy the capability to use this technology or they didn’t have the ability to sequence
0:12:29 everything upfront.
0:12:33 Even if all of the subsequent queries would be technically free later.
0:12:34 Why not?
0:12:35 The way they’re reimbursed.
0:12:36 Oh, how fascinating.
0:12:37 Too expensive, basically.
0:12:38 It’s too expensive.
0:12:43 So even the theoretically there’s an ROI, a return on the investment of sequencing upfront,
0:12:48 just the way the industry structure, the way reimbursement flows, the way payments flow.
0:12:50 It just didn’t make sense for a lot of labs to do this.
0:12:53 So how was that not just a complete roadblock at that point?
0:12:54 It was a big roadblock.
0:12:58 So what that required us to do was to then focus on clinical labs that had the ability
0:13:00 to make certain investments in upfront costs.
0:13:04 And those tended to be very sophisticated labs that do a lot of research work in addition
0:13:08 to patient care and they tended to be on the sort of on the bleeding edge and they wanted
0:13:11 to incorporate new technology and they were great partners and all of that.
0:13:13 But then it goes back to your end of one problem.
0:13:18 So you sell something into that lab and you go next door and next door has a totally different
0:13:21 set of capabilities, a totally different set of constraints, a totally different set
0:13:22 of expectations.
0:13:27 And so therefore, all of a sudden the solution you created for lab A is not relevant or unattainable
0:13:29 for lab B.
0:13:33 Now, to just add to the stepping in it, you know, when you’re analyzing genomic data,
0:13:36 there’s a massive amount of computation required.
0:13:40 And so we went in there assuming, well, this is easy, we’re just going to shoot all of
0:13:44 this up to the cloud, we’ll run the analysis, we’ll send the data back to the lab, the lab
0:13:48 could verify it, generate a report and off we go.
0:13:51 It turns out labs weren’t comfortable sending data up into the cloud, full stop.
0:13:55 At that time, it was just completely– At that time, arguably even today, arguably
0:13:59 even today in 2019, but definitely at that time, we probably should have known that earlier
0:14:03 that would have changed how we thought about going into the clinical lab space.
0:14:04 How would you have done your homework?
0:14:06 I mean, what would that have actually looked like?
0:14:11 It was frankly, I think just defining the specs of what would be required to bring in
0:14:17 our technology, because I think people intuitively know that genomic data is massive, but I don’t
0:14:22 think they know sort of the level of computation required to run the interpretation.
0:14:23 Right.
0:14:24 So like really running the numbers.
0:14:26 Running the numbers for them and by the way, we tried everything.
0:14:30 I mean, we brought representatives from AWS that could show them that they had a HIPAA
0:14:35 compliant cloud that they had received all the certifications and it came back to risk
0:14:36 aversion.
0:14:39 So someone, the lab director, saying like, “Look, I’m sure all of that’s true, but I’m
0:14:41 not going to risk sending all of this data up into the cloud.”
0:14:46 So that was a big, big challenge for us and it ended up being a major limitation for our
0:14:50 ability to expand into the clinical setting because of all of those barriers.
0:14:51 So what did you do?
0:14:56 We had to do a plan A and a plan B. And so the plan A was we assumed that there would
0:15:01 be a couple of forward looking labs or forward thinking labs that would be willing to work
0:15:05 in the cloud environment, much easier to deploy there.
0:15:10 The plan B was we had to create a box and we had to create a box and the box had to have
0:15:11 essentially the competition.
0:15:12 A normal appliance.
0:15:13 Yeah.
0:15:14 We had a normal appliance.
0:15:15 Remember that.
0:15:16 Oh my gosh.
0:15:17 Because they didn’t want the data to go outside.
0:15:21 And it’s for the reasons that we’d expect, you know, there’s regulatory, there’s risk
0:15:24 associated with that today in 2019.
0:15:28 In fact, the companies that have managed to use this technology have taken the sort of
0:15:30 full stack service approach.
0:15:35 So that sort of high low strategy became the approach is get folks to deploy into the cloud
0:15:37 when they were willing to.
0:15:43 And in the case where folks needed an appliance, we basically had to go to labs that had enough
0:15:47 a sample volume that an appliance made sense for them and make basically the case there
0:15:48 from an investment standpoint.
0:15:53 So again, multiple choice, variety and like addressing in different ways.
0:15:58 A pure software company in healthcare is a really hard thing to do.
0:16:01 Because on the one side, you have this challenge that it’s hard to create a sort of a solution
0:16:03 that’s going to fit everyone.
0:16:08 And therefore you need to have some level of services around that software.
0:16:09 That’s on one extreme.
0:16:12 So when you need to have humans in the process or in the loop.
0:16:16 And then the other extreme, if it is pure software, then it’s considered that it should
0:16:17 be free.
0:16:18 So it’s very hard to abstract value.
0:16:19 That’s so interesting.
0:16:23 Do you think that’s shifting at all with the kind of understanding of the importance of
0:16:24 data and some other things?
0:16:25 Yeah.
0:16:27 Look, I would argue it’s shifting on a couple of axes.
0:16:30 The first one is data is becoming more and more valuable.
0:16:36 Historically data was viewed as being either too small in terms of its impact, too narrow,
0:16:38 too dirty, et cetera, et cetera.
0:16:39 Too difficult.
0:16:40 Yeah.
0:16:41 Too unstructured.
0:16:42 So that historically has been the case.
0:16:47 So if you have ways to ingest data and clean it and make it meaningful, then I think that
0:16:48 is valued.
0:16:52 Probably the most public one is what Flatiron was able to do and ultimately getting acquired
0:16:55 by Roche for $2 million.
0:17:00 That’s viewed as using an electronic medical record to capture patient experiences, take
0:17:05 that information, and give researchers the ability to drive valuable insights from that.
0:17:06 That’s a relatively new thing.
0:17:08 So I think there is the ability to create value there.
0:17:10 So I think that’s one axis.
0:17:14 I think there’s a general shift in the model that having a tech-enabled service can be
0:17:18 a valuable thing and if done well can be a scalable business.
0:17:23 In other words, if you know what you’re trying to build and if the software layer reduces
0:17:29 sufficient friction in the system and allows you to add people, not linearly as you scale,
0:17:33 but in a leverageable way, then all of a sudden you could have tech-enabled services that
0:17:35 can grow and become large businesses.
0:17:40 So leaning into what it is that makes it difficult almost and then scaling that, leveraging that.
0:17:41 Exactly.
0:17:42 Finding ways to make that scalable.
0:17:43 Yeah.
0:17:45 That’s not easy to do, but I think it is now doable in a way that probably it wasn’t.
0:17:46 Yeah.
0:17:50 So we see that same trend actually happening in the consumer world where you used to have
0:17:55 a bunch of services like the marketplaces that were purely tech and were just matching
0:17:59 supply and demand and then getting out of the way, whereas now you see a lot more services
0:18:03 like in the real estate market where they’re actually managing properties.
0:18:06 We’re actually going to clean the place and make sure it has good furniture and all that
0:18:07 kind of stuff.
0:18:11 I think the same premise holds true in healthcare where you realize that in order to truly make
0:18:15 an impact, you kind of have to own certain parts of the full stack and that’s what you
0:18:17 see playing out in the rest of the world as well.
0:18:18 Okay.
0:18:22 So we’ve talked about kind of knowing the workflow and the complexity of the system,
0:18:26 running the numbers and specking it out as concretely as possible.
0:18:28 How about in terms of team building?
0:18:33 Are there ways that you, knowing what you knew down the road that you would have changed
0:18:36 how you thought about building the team from the very beginning?
0:18:38 My prior experience was not in healthcare.
0:18:42 And so a lot of my views on how to do these kinds of things were informed by a company
0:18:45 that was just a pure enterprise software company.
0:18:48 And one of the mantras was you want to, in the early stages of a company, hire for all
0:18:52 around athletes and just people who are utility players who can like roll with the punches
0:18:53 and figure it out.
0:18:57 It doesn’t matter what kind of experience they had as long as they’re scrappy, intellectually
0:19:00 motivated people, they’re going to figure it out.
0:19:04 So it certainly took that approach when we started Kyrus and hired folks not necessarily
0:19:09 from healthcare who maybe had some engineering experience or sales experience from elsewhere
0:19:12 in the world and said, “We’re just going to go in there and figure it out.”
0:19:15 But you surely had some deep experts in the space as well.
0:19:17 So my co-founder is a physician by training.
0:19:23 So we had sort of the deep clinical knowledge, but I would say actually we didn’t have that
0:19:28 many people who knew the specific market that we were going after.
0:19:32 And that’s another characteristic of healthcare startups is healthcare is so massive that
0:19:36 when you talk about market segment, you have to be very specific about what you’re talking
0:19:37 about.
0:19:41 So when people come and say, “Oh, I have a company that sells to providers,” I’m like,
0:19:42 that’s great.
0:19:43 That’s like, you know…
0:19:44 What does that actually mean?
0:19:45 Yeah.
0:19:47 Like there’s 20 billion ways that you could describe providers like, “Are you selling
0:19:48 to hospitals?
0:19:49 Are you selling to health systems?
0:19:50 Are you selling to individual practices?”
0:19:53 And each of those can be multi-billion dollar markets in and of themselves.
0:19:56 I used to work in publishing and it reminds me of people who would pitch their books to
0:19:58 us and be like, “It’s for the general reader.”
0:19:59 And you’re like, “Who?
0:20:00 Who?
0:20:01 There is no general reader.”
0:20:02 Exactly.
0:20:03 There’s like somebody who likes to read Amy Tan.
0:20:07 There’s somebody who likes to read like, you know, Dan Brown or whatever.
0:20:08 These are different people.
0:20:09 Yeah.
0:20:10 There you go.
0:20:11 So yeah.
0:20:14 So basically we had folks in our company who had “health care experience,” but maybe
0:20:18 it was from the pharma industry or from payer or even like a different segment of the health
0:20:22 of the provider market, but not the specific market that we were going after, which was
0:20:24 like a very esoteric…
0:20:28 We were going after the biggest health systems, like the top down approach in the enterprise
0:20:29 space.
0:20:32 And there’s very specific characteristics to those organizations that are very different
0:20:34 than even smaller hospital networks.
0:20:38 The areas of the team building exercise that I wish we had been more thoughtful about were,
0:20:43 you know, in terms of customer facing roles, where it was a team responsible for managing
0:20:47 the customer relationship longer term, you know, just how important it is for those people
0:20:54 to have some kind of understanding and empathy and ideally experience with the kind of people
0:20:55 that we were servicing.
0:20:59 There is total merit to saying, “Actually, we need some insiders who might not have any
0:21:05 technical skills whatsoever, but can help us understand the culture and the politics
0:21:08 and what it means to even like talk to a physician.”
0:21:11 You know, we had a bunch of folks who had never been in health care who walked into meetings
0:21:15 and called doctors by their first names, and that was a complete taboo in certain cultures
0:21:17 where you have to call them Dr. Jones or Dr. Smith.
0:21:21 Like “Stranger in a strange land” kind of like, “Here’s the language here.”
0:21:22 Yeah.
0:21:24 So I think from a team building experience, one of the biggest lessons that we certainly
0:21:31 learned was a valuing health care domain expertise earlier in the evolution of a company relative
0:21:36 to other sectors, and then also thinking about where that makes sense, like what functions
0:21:39 that makes sense, because it’s not a 100% universal statement across the board.
0:21:43 I would say our engineering team, it was actually better that they came from outside
0:21:44 of health care.
0:21:48 Oh, so in specific areas of where you need the knowledge and where you don’t.
0:21:51 Why was it a bad thing for engineers to have that?
0:21:53 Not a bad thing per se, but you wanted people who could like really think out of the box
0:21:58 and not be sort of married to the way it’s done today, because actually that’s exactly
0:22:02 the point of building companies in this space is to not do it the way it’s been done.
0:22:07 And so most of the technology systems that are in place are written on super legacy technologies
0:22:10 and don’t have things like APIs and whatnot.
0:22:13 You need to be like super creative about like how to get into these systems and get data
0:22:18 out because they were like fundamentally not designed to have liquidity around the data
0:22:19 that’s stored in them.
0:22:23 And so it was helpful to have people from the financial services industry, for instance,
0:22:27 who had figured those things out with similar banking systems and whatnot and could kind
0:22:30 of bring some of that creativity to the health care space.
0:22:33 So engineering is definitely a space where I felt there was a positive to not having
0:22:37 that health care domain knowledge, but certainly on the commercial side of the business.
0:22:39 I think it’s critically important.
0:22:42 Making sure that the engineering team is as modern as possible is the most valuable thing
0:22:43 you can do for your company.
0:22:48 Because I think what’s generally true and probably definitely true across the board
0:22:52 is that health care, the data sets are so complex, right?
0:22:56 They’re complex in terms of their variety, they’re complex in terms of their volume,
0:22:57 they’re unstructured.
0:22:58 There’s regulatory requirements.
0:23:02 There’s so many things that are challenging from a data handling standpoint.
0:23:06 So building the pipes in the most modern way possible, absolutely critical.
0:23:11 Whoever’s customer facing, I think has to be from that game, has to understand the space,
0:23:15 has to understand who the customer is, has to understand the cultural norms and all of
0:23:16 those things.
0:23:17 Those things are both true.
0:23:22 You need both in the get-go, industry specific on the customer spacing side and domain expert
0:23:24 from the engineering side, right?
0:23:27 And then let’s talk a little bit about the middle, the product, right?
0:23:28 That’s where the sausage gets made.
0:23:29 Totally.
0:23:32 I’m going to be biased because I was the chief product officer of my company.
0:23:35 And that’s where I would say it was split, where I do think it’s important for the leader
0:23:40 of that organization to have a pretty deep understanding of the market.
0:23:44 And so I happen to have had health care experience, not specifically in this particular segment
0:23:48 per se, but I understood some of those cultural nuances and just dynamics of how the market
0:23:51 worked to be able to set strategy.
0:23:55 Below me, however, some of my best product managers were not health care people at all.
0:23:59 And in fact, we had three products, one that was the call center product that I mentioned
0:24:04 earlier, where the end users themselves were not health care people, right?
0:24:09 And so some of them were like high school graduates who go home and they use their iPhone and
0:24:11 they’re used to all these modern technologies and the rest of their lives.
0:24:15 And then they come to work and they’re faced with these totally esoteric, crappy, hard
0:24:17 to use systems.
0:24:20 And so I wanted someone who actually had kind of a consumer mindset.
0:24:24 Did you find yourself doing a lot of sort of explaining and educating though to bridge
0:24:25 that gap?
0:24:26 Yeah.
0:24:28 My philosophy was just throw them in the deep end.
0:24:32 As part of the onboarding experience at Kyrus, you had to visit a hospital call center and
0:24:34 they actually let you listen in on calls.
0:24:37 It was like a religious transformation for these team members who went.
0:24:42 Some came back and said, I cannot believe that this is how these organizations operate,
0:24:43 right?
0:24:45 Cause like everyone thinks of health care as this very pristine, like I’m going to trust
0:24:49 you with my life and they’ll come back and be horrified because, you know, they see that
0:24:54 things are being run on paper and just how much burden they put on the customer, right?
0:24:57 Because part of what you hear when you’re listening in on these calls is like asking
0:24:59 the patient, what do you want to do?
0:25:01 And the patient’s like, well, why would I have understood?
0:25:04 I’m calling you guys a hospital and you’re supposed to tell me what to do.
0:25:05 So that was one reaction.
0:25:07 The other reaction was completely emotional, right?
0:25:11 Because a lot of these patients who were calling in had just been diagnosed with cancer and
0:25:16 they have no idea what they’re doing and they’re calling because they need help.
0:25:20 And then the call center agent sometimes felt helpless because they didn’t have the tools
0:25:21 or the workflows or the information.
0:25:25 Oh, it reminds me of like a 911 operator with like no training, somebody thrown into
0:25:28 the middle of like, I’m having a massive life crisis.
0:25:29 Yeah.
0:25:30 It was inspiring and motivational.
0:25:33 And so that became part of like our training process was to just go out there and see it
0:25:35 versus me explaining it.
0:25:36 That’s really interesting.
0:25:37 Okay.
0:25:38 So what about timing?
0:25:42 Do you think it’s different in the healthcare space, how you think about what’s the right
0:25:43 moment for your product?
0:25:48 One of the big challenges in healthcare is this idea that you can be too early.
0:25:50 You can be too early for a couple of reasons.
0:25:59 One is you need a lot of changes to workflows for the entire system to become much more modern.
0:26:04 But you think this is different from being too early with like pets.com.
0:26:05 That’s a good question.
0:26:09 So the way I would think about it, I described what was for us at the company, a very obvious
0:26:11 evolution of where genetic testing would go.
0:26:15 You would sequence everything first and you would test multiple times in silicone.
0:26:16 You could see the light at the end of the tunnel.
0:26:17 I mean, that’s a clear future.
0:26:22 And so the question is when is the system ready for your particular solution to a problem
0:26:24 that everyone agrees exists, right?
0:26:28 Everyone agrees that we have to do a better job at being able to diagnose folks with genetic
0:26:29 disease.
0:26:33 And I think everyone would agree that using genomics, the ability to do this at large
0:26:38 scale to query multiple times, to use software, to make intelligent queries would be a very
0:26:41 powerful tool, a very powerful solution for that.
0:26:47 But the reality was, continues to be, that just the structures of the industry are such,
0:26:50 even though that’s where I think we will end up, it’s just not ready for it now.
0:26:54 And I think this is true for any entrepreneur, timing is a big part of anything you do.
0:26:58 I think timelines are especially warped in healthcare because it just takes a long time
0:27:00 to adopt new technologies.
0:27:04 There actually is a peer-reviewed study of the average number of years it takes for
0:27:08 new technologies that are introduced into the medical setting to become mass-market
0:27:09 adopted.
0:27:10 Oh, how fascinating.
0:27:11 Wait, wait, let’s guess.
0:27:12 Two years.
0:27:13 17 years.
0:27:14 No!
0:27:15 I mean, we still have fax machines.
0:27:16 That’s true.
0:27:17 We still have fax machines.
0:27:18 We still use the same…
0:27:21 But we’re not talking about when technology leaves, but you’re right.
0:27:22 It’s the same thing, really.
0:27:23 Yeah, so it gets replaced.
0:27:26 Yeah, you can think about it as all the things that have tried to replace the fax machine
0:27:28 or not yet mass-market adopted.
0:27:29 And it’s the same.
0:27:30 You could see it in…
0:27:35 I think the study actually focused primarily on stethoscopes and thermometers and things
0:27:38 that literally have not been redesigned for hundreds of years because it’s been so hard
0:27:39 to disrupt them.
0:27:42 Over the last 17 years, there’s been a bajillion better versions of the stethoscope that we
0:27:43 are just not seeing.
0:27:45 The wheel could have been reinvented, but better.
0:27:46 Absolutely.
0:27:50 Those are the tangible examples, but the same applies to software and technology.
0:27:54 And that’s a lot of the reason why you see the market-leading companies that own the EHR
0:27:57 space today are literally 45 years old.
0:28:01 And by the way, those companies also didn’t hit their stride until like 20 years into
0:28:02 their journeys.
0:28:06 So time functions completely differently, basically, in this system.
0:28:07 It’s almost like…
0:28:08 It’s like a wormhole.
0:28:12 And second of all, it’s an incredible testament to the strength of these systems that…
0:28:13 Totally.
0:28:14 Yeah.
0:28:16 It’s like, once you do make it, it’s totally sticky.
0:28:21 The LTV, essentially, of tech companies that actually make it and get to a certain level
0:28:23 of scale is through the roof.
0:28:25 There’s no incentive to rip them out because if they work, they work.
0:28:29 The switching costs because of all the human and cultural elements that we described is
0:28:30 huge.
0:28:31 Yeah.
0:28:34 So the longevity of your company, if you’re looking at success, is also incredibly promising.
0:28:35 Yeah.
0:28:38 I mean, certainly at Kairis, the way we mitigated it was we thought about what our fundraising
0:28:42 strategy would be to give ourselves enough runway to have that model play out.
0:28:47 We needed to fund the sales cycles and the adoption cycles to create a new category of
0:28:48 solution that didn’t exist.
0:28:51 Did it hang out in the wormhole for a while?
0:28:52 It’s a big oxygen tank.
0:28:53 Yes.
0:28:57 Global happens in healthcare in under three years, and so you kind of have to give it
0:28:58 some runways.
0:29:01 This is one of the things that we’ve spent time talking about is what does a minimal viable
0:29:03 product in healthcare look like?
0:29:04 Doesn’t exist.
0:29:05 Big bang.
0:29:09 You’ve got to go in and you’ve got to create a category and you’ve got to get that adopted.
0:29:14 I think in other industries, you can sort of quote-unquote get away with having a product
0:29:18 that does one thing really, really well and then start there and yes, expand over time,
0:29:23 but at least you can get by and to prove your value with that initial use case.
0:29:26 I think going back to a lot of the points you made earlier in healthcare, when you’re
0:29:32 in the flow of impacting a patient encounter and saying you’re going to rip something out
0:29:36 or change the way that you’re doing something or what have you, you have to make sure that
0:29:39 it’s going to give you the right answer, so to speak.
0:29:43 Even if it’s just one feature, it might mean, okay, yes, it could be one feature, but you
0:29:46 have to be integrated into seven different systems to make sure that the data flowing
0:29:50 into that one feature is enough to inform the right outcome or decision.
0:29:52 So really fully baked.
0:29:55 If a transaction falls through the cracks, while you’re doing some kind of revenue cycle
0:29:59 type encounter, you might not get paid for a procedure that could have a severe impact
0:30:01 on your bottom line.
0:30:02 You need more funding.
0:30:04 You need to think differently about your strategy for product and what that footprint
0:30:05 looks like.
0:30:08 You have to have the full solution.
0:30:12 And the related point I would make to that is it’s really hard to have a point solution.
0:30:14 Even if that point solution is very, very good.
0:30:18 I think people in general in the healthcare system are looking to buy a complete solution.
0:30:23 So if you take the problem from A to B to C to D, that’s great, but they need A to Z.
0:30:25 They can’t get A to Z from you.
0:30:27 It’s very hard to get them to buy A to C from you.
0:30:29 I’ll go even further than Julie.
0:30:32 I will say, not only does MVP not exist in healthcare, I would argue that product-market
0:30:34 fit doesn’t exist in healthcare.
0:30:35 What do you mean by that?
0:30:40 The definition of product-market fit is when the right product meets a good market.
0:30:44 All of the things we talked about creates such distortions in the marketplace that by the
0:30:49 time you actually get through all the hoops, you have such a skewed product.
0:30:51 It’s not really product-market fit.
0:30:53 It’s almost like accepted product capture.
0:30:56 Here you have regulatory issues.
0:30:57 You have pricing concerns.
0:30:58 You have incumbents.
0:31:03 You have so many aspects that distort the market that I would argue that you don’t have a normally
0:31:06 functioning market for software in healthcare.
0:31:12 How would you both embrace that distortion early on and not get completely knocked off
0:31:14 your path by it?
0:31:19 It strikes me that a lot of what you’re describing is know-thyself, know-yourself very deeply.
0:31:22 That was the tagline I know, by the way.
0:31:25 Oh, was it really?
0:31:26 That’s really funny.
0:31:29 I did not work that in for you.
0:31:34 But also know where you’re going and do that deep, I want to say, soul-searching on a company
0:31:37 level and build out accordingly.
0:31:40 How do you get that big center of gravity of really knowing yourself, knowing where
0:31:44 you’re going, but be able to be flexible with that distortion along the way?
0:31:48 The only North Star you can have, and this is going to sound cliche, but really understanding
0:31:54 your value proposition truly from the customer standpoint, it becomes a critical guide for
0:31:55 what you do.
0:31:58 This is a debate that healthcare companies have all the time, which is should your value
0:32:02 proposition be I’m going to save the system money because the healthcare system is very
0:32:05 inefficient and it runs on very low margins generally.
0:32:08 Should it be that I am going to result in better outcomes for patients?
0:32:13 Is it going to be I’m going to create some sort of a lift in terms of return on investment?
0:32:16 There’s a bunch of different ways you can think about value proposition.
0:32:20 If you don’t have that crystal clear from the outset, the amount of obstacles that you
0:32:24 are going to hit along the way are going to make it such that it’s going to be very difficult
0:32:25 to get to the other side.
0:32:29 If you don’t really understand the workflow and the culture and the regulation and the
0:32:33 governance and the politics and all of the other things, you can have a theory on what
0:32:37 the value proposition is, but you need your customer to confirm that early on and sadly
0:32:40 the best way to confirm that is to have them buy something, obviously.
0:32:44 Julie and I have had this debate before, which is a lot of the software platforms that go
0:32:49 into healthcare have been sort of predicated on we’re going to cut costs.
0:32:54 I don’t know of any sort of solution out there that has meaningfully been able to make a
0:32:56 very, very strong case that they can cut costs.
0:32:59 And by the way, part of it is, I think, is because it’s really hard to measure costs.
0:33:03 It’s almost like a necessary evil where you have to say in some way, shape, or form you
0:33:06 are going to reduce costs, but that can’t be your primary value proposition.
0:33:09 Because at the end of the day, it’s aligned in the cost structure that can get wiped out
0:33:13 over time and potentially get commoditized.
0:33:17 So is the takeaway, know your value proposition as early as possible and test it?
0:33:20 That and then have the conversation of like, okay, if we are able to accomplish what we
0:33:23 just described, is it worth it?
0:33:24 Is the juice worth the squeeze?
0:33:29 Because it’s so expensive to distribute product in this market because of the sales cycles
0:33:34 and the nature of the enterprise sales motion and whatnot, that if you’re not able to envision
0:33:39 a path towards being like at least a half a million dollar kind of a year type solution
0:33:42 in this space, it’s actually not financially worth it to build a business in that area.
0:33:43 Right.
0:33:45 Which goes back to your point of like run the numbers, basically.
0:33:48 At least like back at the envelope, like, you know, whiteboard kind of thing.
0:33:52 I mean, is there anything that you can figure out as you go?
0:33:57 It sounds like you need to know so much before you begin and be so self aware and so kind
0:33:59 of like have the end game in sight.
0:34:03 Are there things that you can leave sort of more organic and like feel out as you go?
0:34:04 Yeah.
0:34:05 No, I mean, absolutely.
0:34:08 There’s tons of things you can be doing on a daily basis with end users and just like
0:34:12 feedback mechanisms on like how people are, are they actually able to do their jobs, for
0:34:15 instance, and making minor tweaks to the workflows and whatnot.
0:34:19 So that was always, you know, a component of a more organic and dynamic aspect of how
0:34:20 we did things.
0:34:24 The other thing that you need to kind of think about doing in parallel is, you know, so much
0:34:29 of success of technology and healthcare is predicated on integrating into other ecosystem
0:34:30 players.
0:34:33 And so this is actually probably one industry where you definitely can’t like just build
0:34:34 in a vacuum.
0:34:38 You actually should understand, even if it’s not, you know, for another few years that you’re
0:34:42 really going to have to do this, like who are the players, we just need to get to know
0:34:46 so that we’re on their radar when time comes for us to take the hammer and like try to break
0:34:50 down the wall of integration with that vendor that we are on their good side and that they
0:34:53 know who we are so we can kind of make that happen faster.
0:34:58 So things like that, I think you can be doing in parallel to, you know, the kind of formulation
0:35:00 of what the footprint of the product is.
0:35:04 If you’ve got the right solution, you can get very creative in how you get paid.
0:35:09 So figuring out different pricing structures or value capture mechanisms, I think is something
0:35:13 that you can do pretty organically because if you are making a difference in the system,
0:35:17 the system has so much cost built into it and so much revenue flowing through it that
0:35:19 there are ways to be very imaginative there.
0:35:21 So that’s the first thing I would say.
0:35:26 The second thing I would say is thinking about adjacencies, you know, going from one, you
0:35:32 know, your core function to the next adjacent use case, not all adjacencies are created
0:35:33 equal.
0:35:34 One might be easier than the other.
0:35:38 It’s almost like, you know, jumping on stones across a pond or something, right?
0:35:41 What’s the next stone I can jump on that’s least likely to make me fall into the water,
0:35:42 right?
0:35:43 Yeah.
0:35:44 Even if it doesn’t get me as far as another one.
0:35:45 Right.
0:35:46 Always have that closer spot insight.
0:35:49 You’re almost, you’re creating the next thing and the next thing and the next thing and
0:35:50 you build out from there.
0:35:54 And eventually you cover so much surface area that, you know, you become a very sticky solution
0:35:58 and you hopefully become a complete solution sort of closer to the A to Z type vision.
0:35:59 Okay.
0:36:00 Last question.
0:36:01 Biggest takeaways.
0:36:04 Quick lightning round for your founder struggling right now.
0:36:05 What would you say?
0:36:06 Bullet points.
0:36:07 Know your market segment.
0:36:12 Be very specific about what segment you’re going after because that has major implications
0:36:14 for your go to market and your product.
0:36:15 Good one.
0:36:16 All right.
0:36:17 Let me get that to you.
0:36:20 One is build the multidisciplinary team early.
0:36:24 Two is understanding and if the person that suffers from the pain point can actually pay
0:36:29 for your solution because there’s a lot of misincentives in the healthcare system.
0:36:34 And three, with the right technology, you can have massive impact on patient lives and
0:36:38 the experience that we have with the healthcare system, which we will all touch in our lifetime.
0:36:42 And if there’s anything you can do to make it better as an entrepreneur, I would say
0:36:44 that is extraordinarily satisfying.
0:36:45 That’s fantastic.
0:36:46 And good bullets.
0:36:49 Thank you both so much for joining us on the A16Z podcast.
0:36:50 Thank you.
0:00:06 I’m Hannah, and this episode is all about building a software company in healthcare.
0:00:11 In this conversation, Jorge Conde, A16Z general partner in bio and healthcare, previous founder
0:00:16 of the genomics company Nome, and Julie Yu, partner on the deal team for the Bio Fund,
0:00:21 and previous founder of the patient provider matching system Kyrus, explained what it is
0:00:25 that makes building a company in the healthcare space so fundamentally different from in other
0:00:26 sectors.
0:00:28 And why exactly it’s so damn hard.
0:00:32 So let’s start with basically just the very fundamental difference between building a
0:00:36 software company full stop and building a software company in the healthcare space.
0:00:39 What are the most foundational, crucial differences?
0:00:45 Well, historically at least, software had two very important sort of qualities in healthcare.
0:00:48 The first one, the actual quality of software deployed in healthcare system historically
0:00:49 has not been great.
0:00:51 User interface wise and experience wise.
0:00:52 Bad track record.
0:00:53 Bad track record there.
0:00:56 And the second one is that it was usually not highly valued.
0:01:01 So at least a lot of times it was considered either free or cheap.
0:01:02 And why was that?
0:01:05 That in it from the very beginning, there was not a lot of value attached to this.
0:01:09 On the healthcare system, a lot of things still have a very human component to them,
0:01:14 automating things and sort of creating frictionless experiences or delightful experiences.
0:01:17 The things that software is really good at doing, it’s just really hard to do in the
0:01:18 healthcare system.
0:01:21 The second one is, I’m going to generalize for a second, but I think a lot of times in
0:01:26 the healthcare system, software has sold us a component of a broader service or of a
0:01:27 broader offering.
0:01:31 And so therefore it’s the piece that tends to get sort of devalued first because it obviously
0:01:32 has the lowest marginal cost.
0:01:36 It’s going to create this weird dynamic for software companies that are trying to build
0:01:37 in healthcare.
0:01:41 There’s a higher degree of sensitivity in this particular market for things that get
0:01:44 in the way of the patient provider experience.
0:01:47 One of the challenges/opportunities within healthcare is that it tends to be much more
0:01:50 risk averse when it comes to adoption of new technologies.
0:01:55 One meaningful difference in introducing a software product to this market versus other
0:02:02 markets is the level of scrutiny and the bar that you need to hit from a not even usability
0:02:07 perspective but just utility and actually having validation of if you are going to introduce
0:02:13 something new into the care delivery flow, it better work because the stakes are so high.
0:02:16 If you get it wrong, you could either send a patient in the wrong direction or they might
0:02:20 not get the care that they need or it could actually harm the individuals involved.
0:02:23 So not just higher barrier to entry but higher stakes, correct?
0:02:24 Immediately.
0:02:26 They’re a reticent buyer, generally speaking.
0:02:30 They’re running on very thin margins if we’re selling into the healthcare system, into provider
0:02:37 space and it needs to work because if it doesn’t, obviously there can be patient harm so the
0:02:42 probability that a newcomer, an upstart can come in and make that case in a convincing
0:02:44 way is a very, very difficult challenge.
0:02:48 So does that mean you have to have certain prerequisites that you may not need to have
0:02:49 in other spaces?
0:02:52 If you know you have these challenges and you know that you’re entering this space with
0:02:55 a lot more barrier to entry and a lot higher stakes, are there certain things you need
0:03:01 in place, a certain kind of proof of concept that you might not have to have otherwise?
0:03:04 Well, first of all, I think you’re touching on a very important thing which is in the space
0:03:07 and I’m going to specifically focus on sort of the healthcare system.
0:03:11 So let’s call it provider systems, payers and the like.
0:03:17 You have to really understand what the workflows are, what the problem space is and how to
0:03:19 actually address any of those things.
0:03:24 And so one of the biggest challenges I think that companies have when they want to build
0:03:28 software products here is to really understand what problem they’re going to solve because
0:03:33 I think you have this weird sort of intersection between it’s very non-intuitive, it’s still
0:03:39 very human driven and centric, there are regulatory barriers, you don’t want to get in between
0:03:42 say a provider and a patient, you know, most people aren’t born with the ability to say
0:03:48 like I know I can insert a piece of software into this part of the workflow and I will
0:03:50 solve an acute pain point for the system.
0:03:51 That’s not obvious.
0:03:53 And some of that is actually lack of standardization.
0:03:57 You would think that medicine is an industry that has a tremendous amount of standardization
0:04:01 and protocols around how people make decisions and do things.
0:04:04 But it actually turns out that healthcare is an industry that actually is characterized
0:04:06 by a tremendous amount of variation.
0:04:07 And variation in what kinds of ways?
0:04:11 It could be variation in terms of actually literally the decision that if you have ten
0:04:16 doctors who are all presented with the same patient, you might see ten different decisions
0:04:18 about how to treat that patient.
0:04:22 Some physicians might be more aggressive about using invasive surgical techniques versus
0:04:27 others who are more holistic, even just how I was brought up religiously or culturally
0:04:29 might impact the way I think about that problem.
0:04:33 From a product perspective, you could have multiple drugs that all treat the same condition,
0:04:35 that all have different implications and whatnot.
0:04:39 So even there, even though you have a patient population that is characterized by the same
0:04:43 diagnosis, you could have dozens of different ways that those patients play out.
0:04:47 And so it makes it very hard for a technology company to come in and sort of generalize
0:04:53 and say, you know, there is one single method for, you know, manufacturing this thing or
0:04:57 for making this decision and managing this patient population, ultimately that reflects
0:05:01 as differences in the financial profile of different patients.
0:05:04 Healthcare, it’s like politics, it’s very local.
0:05:09 Thinking that you’re going to have an out-of-the-box, one and done solution, even in systems that
0:05:14 look similar from either a size standpoint or reach standpoint or even a geographic standpoint,
0:05:16 these are all kind of end-of-ones.
0:05:18 So what does that mean?
0:05:24 So we have kind of knowledge of workflow, the knowledge of variety and spectrum and that
0:05:27 you are ultimately working in weirdly an N=1 scenario.
0:05:31 I want to bring it back to like actual practicalities of this sort of company building.
0:05:38 In your experiences, you both founded companies, what do you wish you had known or done differently
0:05:42 from the very beginning, given the complexity of that space and the unique challenges that
0:05:44 building a company in healthcare presents?
0:05:48 With Kairis, one of the products that we had was a product that was used by call center
0:05:50 agents in hospitals.
0:05:53 And our thesis when we first launched the product was, oh, well, we’re just going to
0:05:58 go after every hospital that has a call center and they probably all operate similarly.
0:06:02 And what constitutes the job of a call center agent is probably relatively homogenous.
0:06:06 And so we can make all sorts of assumptions about how it’s built, how it’s deployed and
0:06:07 how it’s managed over time.
0:06:11 The thing that strikes me already is that feels like a reasonable assessment of the
0:06:12 lay of the land.
0:06:13 Yeah.
0:06:15 And especially, I think it’s very easy to get fooled in healthcare by looking at other
0:06:20 industries and seeing how it works in the rest of the world because certainly…
0:06:21 And then you pull up the…
0:06:22 Yeah.
0:06:24 And then you pull up the wool and it’s like, oh, it’s completely the opposite.
0:06:25 Call centers.
0:06:28 I mean, that’s definitely an industry that if you look at retail or even all the airline
0:06:33 companies and how they operate their customer service operations tend to be pretty standardized
0:06:36 and pretty sophisticated in a lot of cases.
0:06:39 When did you start to realize this wasn’t maybe your average call center?
0:06:45 Like on day one, first of all, there’s heterogeneity in the actual scope of services of pretty
0:06:47 much every call center that we encountered.
0:06:51 Some call centers might be fully centralized and they’re like a central 800 number that
0:06:56 receives every call that comes into the hospital versus others that are decentralized that
0:07:01 only serve the primary care line versus the cardiology line versus the dermatology line.
0:07:05 And because of that, they will have just fundamentally different starting points of where they have
0:07:07 to be in the workflow for the thing to work.
0:07:12 The other aspect is the scope of functions that the call center plays.
0:07:15 It could be everything from just a general marketing service where a customer might call
0:07:18 in and say, do you provide these kinds of services?
0:07:20 Can you give me directions to the clinic?
0:07:23 All the way to I need a prescription refill.
0:07:24 I’ve been diagnosed with this thing.
0:07:27 I need to figure out what kind of surgery I need.
0:07:29 So again, much bigger range of possibilities.
0:07:30 Correct.
0:07:31 Yeah.
0:07:36 Like I’m a call center agent and how do you define in my job so that when I give you another
0:07:41 piece of software to use to do that job, it’s going to be seamless.
0:07:44 And when you have that kind of heterogeneity around even the sheer definition of what the
0:07:48 job is, it makes it very hard to design a scalable solution that can kind of fit into
0:07:50 all those different environments.
0:07:55 So day one, we actually were fortunate to get a customer that did have a pretty robust
0:08:00 centralized call center group that was hundreds of people who literally were answering every
0:08:02 call that was coming into the health system.
0:08:06 And so the immediate sort of leap that we made was, oh, they must all look like this.
0:08:11 Even if 80% of it was the same and there was 20% sort of buffer that needed to be modified,
0:08:12 we can deal with that.
0:08:16 Yes, they all had central call centers, but the fundamental scope of jobs that they were
0:08:19 doing were completely different across the board.
0:08:22 And some were more clinical in nature, some were more marketing in nature, some were more
0:08:24 financial in nature, et cetera.
0:08:26 So what were the knock on effects of that?
0:08:27 Yeah.
0:08:32 The impact on like, go to market, product design and spend product strategy.
0:08:36 Most importantly, the service model of you could either say, we’re going to design our
0:08:40 software to be so flexible that it could work in any environment.
0:08:46 Or you could say, we’re going to provide services to come train your people to behave in a more
0:08:49 standardized way relative to the rest of our book of business.
0:08:52 And so we ultimately ended up taking a hybrid approach to both.
0:08:55 But the latter, you know, that services approach is something that we hadn’t thought about
0:09:00 that allowed us to sort of abstract out the variation to some degree, but also provide
0:09:04 value back to the customers in a pretty unique way because then we had the best practices
0:09:07 for, you know, how it should work.
0:09:10 So ultimately it was a good thing, but it was a major fork in the road.
0:09:11 Absolutely.
0:09:15 Because there is so much variability, because there’s so much localization, the notion of
0:09:20 the pure SaaS model where you’re just throwing technology over the fence and assuming that
0:09:25 it will fit into whatever environment you’re deploying it into, that is a moot point in
0:09:30 healthcare, you actually do need to think about the services component of things.
0:09:34 There was a whole generation of companies that got started like a decade ago that took
0:09:39 these sort of tech-only approaches and failed to get scale or had to fundamentally pivot
0:09:43 their models to actually take into account more of the human element of the service delivery
0:09:44 model.
0:09:45 I mean, even there’s a term for it now, right?
0:09:50 Tech-enabled services is a way of doing things now in digital health that I think is well
0:09:56 recognized that it’s necessary to wrap the technology with a human component to essentially
0:10:00 address and be able to accommodate all the variation that you see across different customer
0:10:01 bases.
0:10:04 And it changes your cost structure fundamentally, the nature of how we talked about the business
0:10:05 and how it scales.
0:10:08 And even our fundraising strategy fundamentally changed because of that.
0:10:13 And so we did have to, you know, raise more and give ourselves more runway and think about
0:10:15 different ways to manage our margin.
0:10:19 It sounds like everything that could have been changed by that.
0:10:23 Let’s go back to a specific example where you really put your foot in it.
0:10:26 Well, so in our experience at Noam, it was interesting because here, this is a company
0:10:32 with, the sole purpose of the company was to provide software capability to analyze
0:10:33 genomic information.
0:10:36 And so, you know, when you launch that, your assumption is, well, this could be used to
0:10:39 power all kinds of applications.
0:10:43 It could be used for research, either an academia and industry, it can be used for, you know,
0:10:44 clinical diagnostics.
0:10:45 Flexible.
0:10:46 We thought it was very flexible.
0:10:50 But challenge one is, you know, a solution looking for a problem is always a very, very
0:10:51 dangerous thing.
0:10:52 I think that’s universally true.
0:10:54 I think it’s especially true in the healthcare space.
0:11:00 And challenge two was understanding exactly where, in the case of the clinical setting,
0:11:03 where this technology would be used in the workflow.
0:11:07 So here we wanted to go after the clinical labs.
0:11:08 That was your initial hypothesis?
0:11:12 Our initial hypothesis for an application in a clinical setting.
0:11:17 You have technicians and docs that are inside of the laboratory setting, receiving samples,
0:11:23 running a test, analyzing the results of that test, generating a report that gets signed
0:11:25 off by a lab director that goes back to a physician.
0:11:27 Usually it’s in the form of a diagnosis, right?
0:11:29 And it gets signed off and it goes to the physician.
0:11:36 The physician now takes that report and basically decides what to do based on that information.
0:11:42 So our assumption was, well, if you have the ability to sequence DNA now in a way that
0:11:46 you couldn’t before, before you’d have to do all of these specific tests, you have to
0:11:49 know what to test and then you’d test it and then you’d get a report.
0:11:53 You had to know what street lamp the keys were under, right, like there in that case.
0:11:57 Whereas once you had the full genome, you could just sequence everything and just run
0:11:59 a bunch of software queries.
0:12:04 So our thought going into this was, well, that’s an incredibly powerful tool for clinical labs
0:12:08 because first of all, you can sequence just once and analyze over time.
0:12:09 Right.
0:12:12 It seems like a totally legitimate assumption to make.
0:12:15 And it turns out that there was a lot of challenges with that assumption.
0:12:17 The first one is every lab is different.
0:12:21 A lot of them didn’t have the budget or the willingness to basically pay the upfront
0:12:28 piece to buy the capability to use this technology or they didn’t have the ability to sequence
0:12:29 everything upfront.
0:12:33 Even if all of the subsequent queries would be technically free later.
0:12:34 Why not?
0:12:35 The way they’re reimbursed.
0:12:36 Oh, how fascinating.
0:12:37 Too expensive, basically.
0:12:38 It’s too expensive.
0:12:43 So even the theoretically there’s an ROI, a return on the investment of sequencing upfront,
0:12:48 just the way the industry structure, the way reimbursement flows, the way payments flow.
0:12:50 It just didn’t make sense for a lot of labs to do this.
0:12:53 So how was that not just a complete roadblock at that point?
0:12:54 It was a big roadblock.
0:12:58 So what that required us to do was to then focus on clinical labs that had the ability
0:13:00 to make certain investments in upfront costs.
0:13:04 And those tended to be very sophisticated labs that do a lot of research work in addition
0:13:08 to patient care and they tended to be on the sort of on the bleeding edge and they wanted
0:13:11 to incorporate new technology and they were great partners and all of that.
0:13:13 But then it goes back to your end of one problem.
0:13:18 So you sell something into that lab and you go next door and next door has a totally different
0:13:21 set of capabilities, a totally different set of constraints, a totally different set
0:13:22 of expectations.
0:13:27 And so therefore, all of a sudden the solution you created for lab A is not relevant or unattainable
0:13:29 for lab B.
0:13:33 Now, to just add to the stepping in it, you know, when you’re analyzing genomic data,
0:13:36 there’s a massive amount of computation required.
0:13:40 And so we went in there assuming, well, this is easy, we’re just going to shoot all of
0:13:44 this up to the cloud, we’ll run the analysis, we’ll send the data back to the lab, the lab
0:13:48 could verify it, generate a report and off we go.
0:13:51 It turns out labs weren’t comfortable sending data up into the cloud, full stop.
0:13:55 At that time, it was just completely– At that time, arguably even today, arguably
0:13:59 even today in 2019, but definitely at that time, we probably should have known that earlier
0:14:03 that would have changed how we thought about going into the clinical lab space.
0:14:04 How would you have done your homework?
0:14:06 I mean, what would that have actually looked like?
0:14:11 It was frankly, I think just defining the specs of what would be required to bring in
0:14:17 our technology, because I think people intuitively know that genomic data is massive, but I don’t
0:14:22 think they know sort of the level of computation required to run the interpretation.
0:14:23 Right.
0:14:24 So like really running the numbers.
0:14:26 Running the numbers for them and by the way, we tried everything.
0:14:30 I mean, we brought representatives from AWS that could show them that they had a HIPAA
0:14:35 compliant cloud that they had received all the certifications and it came back to risk
0:14:36 aversion.
0:14:39 So someone, the lab director, saying like, “Look, I’m sure all of that’s true, but I’m
0:14:41 not going to risk sending all of this data up into the cloud.”
0:14:46 So that was a big, big challenge for us and it ended up being a major limitation for our
0:14:50 ability to expand into the clinical setting because of all of those barriers.
0:14:51 So what did you do?
0:14:56 We had to do a plan A and a plan B. And so the plan A was we assumed that there would
0:15:01 be a couple of forward looking labs or forward thinking labs that would be willing to work
0:15:05 in the cloud environment, much easier to deploy there.
0:15:10 The plan B was we had to create a box and we had to create a box and the box had to have
0:15:11 essentially the competition.
0:15:12 A normal appliance.
0:15:13 Yeah.
0:15:14 We had a normal appliance.
0:15:15 Remember that.
0:15:16 Oh my gosh.
0:15:17 Because they didn’t want the data to go outside.
0:15:21 And it’s for the reasons that we’d expect, you know, there’s regulatory, there’s risk
0:15:24 associated with that today in 2019.
0:15:28 In fact, the companies that have managed to use this technology have taken the sort of
0:15:30 full stack service approach.
0:15:35 So that sort of high low strategy became the approach is get folks to deploy into the cloud
0:15:37 when they were willing to.
0:15:43 And in the case where folks needed an appliance, we basically had to go to labs that had enough
0:15:47 a sample volume that an appliance made sense for them and make basically the case there
0:15:48 from an investment standpoint.
0:15:53 So again, multiple choice, variety and like addressing in different ways.
0:15:58 A pure software company in healthcare is a really hard thing to do.
0:16:01 Because on the one side, you have this challenge that it’s hard to create a sort of a solution
0:16:03 that’s going to fit everyone.
0:16:08 And therefore you need to have some level of services around that software.
0:16:09 That’s on one extreme.
0:16:12 So when you need to have humans in the process or in the loop.
0:16:16 And then the other extreme, if it is pure software, then it’s considered that it should
0:16:17 be free.
0:16:18 So it’s very hard to abstract value.
0:16:19 That’s so interesting.
0:16:23 Do you think that’s shifting at all with the kind of understanding of the importance of
0:16:24 data and some other things?
0:16:25 Yeah.
0:16:27 Look, I would argue it’s shifting on a couple of axes.
0:16:30 The first one is data is becoming more and more valuable.
0:16:36 Historically data was viewed as being either too small in terms of its impact, too narrow,
0:16:38 too dirty, et cetera, et cetera.
0:16:39 Too difficult.
0:16:40 Yeah.
0:16:41 Too unstructured.
0:16:42 So that historically has been the case.
0:16:47 So if you have ways to ingest data and clean it and make it meaningful, then I think that
0:16:48 is valued.
0:16:52 Probably the most public one is what Flatiron was able to do and ultimately getting acquired
0:16:55 by Roche for $2 million.
0:17:00 That’s viewed as using an electronic medical record to capture patient experiences, take
0:17:05 that information, and give researchers the ability to drive valuable insights from that.
0:17:06 That’s a relatively new thing.
0:17:08 So I think there is the ability to create value there.
0:17:10 So I think that’s one axis.
0:17:14 I think there’s a general shift in the model that having a tech-enabled service can be
0:17:18 a valuable thing and if done well can be a scalable business.
0:17:23 In other words, if you know what you’re trying to build and if the software layer reduces
0:17:29 sufficient friction in the system and allows you to add people, not linearly as you scale,
0:17:33 but in a leverageable way, then all of a sudden you could have tech-enabled services that
0:17:35 can grow and become large businesses.
0:17:40 So leaning into what it is that makes it difficult almost and then scaling that, leveraging that.
0:17:41 Exactly.
0:17:42 Finding ways to make that scalable.
0:17:43 Yeah.
0:17:45 That’s not easy to do, but I think it is now doable in a way that probably it wasn’t.
0:17:46 Yeah.
0:17:50 So we see that same trend actually happening in the consumer world where you used to have
0:17:55 a bunch of services like the marketplaces that were purely tech and were just matching
0:17:59 supply and demand and then getting out of the way, whereas now you see a lot more services
0:18:03 like in the real estate market where they’re actually managing properties.
0:18:06 We’re actually going to clean the place and make sure it has good furniture and all that
0:18:07 kind of stuff.
0:18:11 I think the same premise holds true in healthcare where you realize that in order to truly make
0:18:15 an impact, you kind of have to own certain parts of the full stack and that’s what you
0:18:17 see playing out in the rest of the world as well.
0:18:18 Okay.
0:18:22 So we’ve talked about kind of knowing the workflow and the complexity of the system,
0:18:26 running the numbers and specking it out as concretely as possible.
0:18:28 How about in terms of team building?
0:18:33 Are there ways that you, knowing what you knew down the road that you would have changed
0:18:36 how you thought about building the team from the very beginning?
0:18:38 My prior experience was not in healthcare.
0:18:42 And so a lot of my views on how to do these kinds of things were informed by a company
0:18:45 that was just a pure enterprise software company.
0:18:48 And one of the mantras was you want to, in the early stages of a company, hire for all
0:18:52 around athletes and just people who are utility players who can like roll with the punches
0:18:53 and figure it out.
0:18:57 It doesn’t matter what kind of experience they had as long as they’re scrappy, intellectually
0:19:00 motivated people, they’re going to figure it out.
0:19:04 So it certainly took that approach when we started Kyrus and hired folks not necessarily
0:19:09 from healthcare who maybe had some engineering experience or sales experience from elsewhere
0:19:12 in the world and said, “We’re just going to go in there and figure it out.”
0:19:15 But you surely had some deep experts in the space as well.
0:19:17 So my co-founder is a physician by training.
0:19:23 So we had sort of the deep clinical knowledge, but I would say actually we didn’t have that
0:19:28 many people who knew the specific market that we were going after.
0:19:32 And that’s another characteristic of healthcare startups is healthcare is so massive that
0:19:36 when you talk about market segment, you have to be very specific about what you’re talking
0:19:37 about.
0:19:41 So when people come and say, “Oh, I have a company that sells to providers,” I’m like,
0:19:42 that’s great.
0:19:43 That’s like, you know…
0:19:44 What does that actually mean?
0:19:45 Yeah.
0:19:47 Like there’s 20 billion ways that you could describe providers like, “Are you selling
0:19:48 to hospitals?
0:19:49 Are you selling to health systems?
0:19:50 Are you selling to individual practices?”
0:19:53 And each of those can be multi-billion dollar markets in and of themselves.
0:19:56 I used to work in publishing and it reminds me of people who would pitch their books to
0:19:58 us and be like, “It’s for the general reader.”
0:19:59 And you’re like, “Who?
0:20:00 Who?
0:20:01 There is no general reader.”
0:20:02 Exactly.
0:20:03 There’s like somebody who likes to read Amy Tan.
0:20:07 There’s somebody who likes to read like, you know, Dan Brown or whatever.
0:20:08 These are different people.
0:20:09 Yeah.
0:20:10 There you go.
0:20:11 So yeah.
0:20:14 So basically we had folks in our company who had “health care experience,” but maybe
0:20:18 it was from the pharma industry or from payer or even like a different segment of the health
0:20:22 of the provider market, but not the specific market that we were going after, which was
0:20:24 like a very esoteric…
0:20:28 We were going after the biggest health systems, like the top down approach in the enterprise
0:20:29 space.
0:20:32 And there’s very specific characteristics to those organizations that are very different
0:20:34 than even smaller hospital networks.
0:20:38 The areas of the team building exercise that I wish we had been more thoughtful about were,
0:20:43 you know, in terms of customer facing roles, where it was a team responsible for managing
0:20:47 the customer relationship longer term, you know, just how important it is for those people
0:20:54 to have some kind of understanding and empathy and ideally experience with the kind of people
0:20:55 that we were servicing.
0:20:59 There is total merit to saying, “Actually, we need some insiders who might not have any
0:21:05 technical skills whatsoever, but can help us understand the culture and the politics
0:21:08 and what it means to even like talk to a physician.”
0:21:11 You know, we had a bunch of folks who had never been in health care who walked into meetings
0:21:15 and called doctors by their first names, and that was a complete taboo in certain cultures
0:21:17 where you have to call them Dr. Jones or Dr. Smith.
0:21:21 Like “Stranger in a strange land” kind of like, “Here’s the language here.”
0:21:22 Yeah.
0:21:24 So I think from a team building experience, one of the biggest lessons that we certainly
0:21:31 learned was a valuing health care domain expertise earlier in the evolution of a company relative
0:21:36 to other sectors, and then also thinking about where that makes sense, like what functions
0:21:39 that makes sense, because it’s not a 100% universal statement across the board.
0:21:43 I would say our engineering team, it was actually better that they came from outside
0:21:44 of health care.
0:21:48 Oh, so in specific areas of where you need the knowledge and where you don’t.
0:21:51 Why was it a bad thing for engineers to have that?
0:21:53 Not a bad thing per se, but you wanted people who could like really think out of the box
0:21:58 and not be sort of married to the way it’s done today, because actually that’s exactly
0:22:02 the point of building companies in this space is to not do it the way it’s been done.
0:22:07 And so most of the technology systems that are in place are written on super legacy technologies
0:22:10 and don’t have things like APIs and whatnot.
0:22:13 You need to be like super creative about like how to get into these systems and get data
0:22:18 out because they were like fundamentally not designed to have liquidity around the data
0:22:19 that’s stored in them.
0:22:23 And so it was helpful to have people from the financial services industry, for instance,
0:22:27 who had figured those things out with similar banking systems and whatnot and could kind
0:22:30 of bring some of that creativity to the health care space.
0:22:33 So engineering is definitely a space where I felt there was a positive to not having
0:22:37 that health care domain knowledge, but certainly on the commercial side of the business.
0:22:39 I think it’s critically important.
0:22:42 Making sure that the engineering team is as modern as possible is the most valuable thing
0:22:43 you can do for your company.
0:22:48 Because I think what’s generally true and probably definitely true across the board
0:22:52 is that health care, the data sets are so complex, right?
0:22:56 They’re complex in terms of their variety, they’re complex in terms of their volume,
0:22:57 they’re unstructured.
0:22:58 There’s regulatory requirements.
0:23:02 There’s so many things that are challenging from a data handling standpoint.
0:23:06 So building the pipes in the most modern way possible, absolutely critical.
0:23:11 Whoever’s customer facing, I think has to be from that game, has to understand the space,
0:23:15 has to understand who the customer is, has to understand the cultural norms and all of
0:23:16 those things.
0:23:17 Those things are both true.
0:23:22 You need both in the get-go, industry specific on the customer spacing side and domain expert
0:23:24 from the engineering side, right?
0:23:27 And then let’s talk a little bit about the middle, the product, right?
0:23:28 That’s where the sausage gets made.
0:23:29 Totally.
0:23:32 I’m going to be biased because I was the chief product officer of my company.
0:23:35 And that’s where I would say it was split, where I do think it’s important for the leader
0:23:40 of that organization to have a pretty deep understanding of the market.
0:23:44 And so I happen to have had health care experience, not specifically in this particular segment
0:23:48 per se, but I understood some of those cultural nuances and just dynamics of how the market
0:23:51 worked to be able to set strategy.
0:23:55 Below me, however, some of my best product managers were not health care people at all.
0:23:59 And in fact, we had three products, one that was the call center product that I mentioned
0:24:04 earlier, where the end users themselves were not health care people, right?
0:24:09 And so some of them were like high school graduates who go home and they use their iPhone and
0:24:11 they’re used to all these modern technologies and the rest of their lives.
0:24:15 And then they come to work and they’re faced with these totally esoteric, crappy, hard
0:24:17 to use systems.
0:24:20 And so I wanted someone who actually had kind of a consumer mindset.
0:24:24 Did you find yourself doing a lot of sort of explaining and educating though to bridge
0:24:25 that gap?
0:24:26 Yeah.
0:24:28 My philosophy was just throw them in the deep end.
0:24:32 As part of the onboarding experience at Kyrus, you had to visit a hospital call center and
0:24:34 they actually let you listen in on calls.
0:24:37 It was like a religious transformation for these team members who went.
0:24:42 Some came back and said, I cannot believe that this is how these organizations operate,
0:24:43 right?
0:24:45 Cause like everyone thinks of health care as this very pristine, like I’m going to trust
0:24:49 you with my life and they’ll come back and be horrified because, you know, they see that
0:24:54 things are being run on paper and just how much burden they put on the customer, right?
0:24:57 Because part of what you hear when you’re listening in on these calls is like asking
0:24:59 the patient, what do you want to do?
0:25:01 And the patient’s like, well, why would I have understood?
0:25:04 I’m calling you guys a hospital and you’re supposed to tell me what to do.
0:25:05 So that was one reaction.
0:25:07 The other reaction was completely emotional, right?
0:25:11 Because a lot of these patients who were calling in had just been diagnosed with cancer and
0:25:16 they have no idea what they’re doing and they’re calling because they need help.
0:25:20 And then the call center agent sometimes felt helpless because they didn’t have the tools
0:25:21 or the workflows or the information.
0:25:25 Oh, it reminds me of like a 911 operator with like no training, somebody thrown into
0:25:28 the middle of like, I’m having a massive life crisis.
0:25:29 Yeah.
0:25:30 It was inspiring and motivational.
0:25:33 And so that became part of like our training process was to just go out there and see it
0:25:35 versus me explaining it.
0:25:36 That’s really interesting.
0:25:37 Okay.
0:25:38 So what about timing?
0:25:42 Do you think it’s different in the healthcare space, how you think about what’s the right
0:25:43 moment for your product?
0:25:48 One of the big challenges in healthcare is this idea that you can be too early.
0:25:50 You can be too early for a couple of reasons.
0:25:59 One is you need a lot of changes to workflows for the entire system to become much more modern.
0:26:04 But you think this is different from being too early with like pets.com.
0:26:05 That’s a good question.
0:26:09 So the way I would think about it, I described what was for us at the company, a very obvious
0:26:11 evolution of where genetic testing would go.
0:26:15 You would sequence everything first and you would test multiple times in silicone.
0:26:16 You could see the light at the end of the tunnel.
0:26:17 I mean, that’s a clear future.
0:26:22 And so the question is when is the system ready for your particular solution to a problem
0:26:24 that everyone agrees exists, right?
0:26:28 Everyone agrees that we have to do a better job at being able to diagnose folks with genetic
0:26:29 disease.
0:26:33 And I think everyone would agree that using genomics, the ability to do this at large
0:26:38 scale to query multiple times, to use software, to make intelligent queries would be a very
0:26:41 powerful tool, a very powerful solution for that.
0:26:47 But the reality was, continues to be, that just the structures of the industry are such,
0:26:50 even though that’s where I think we will end up, it’s just not ready for it now.
0:26:54 And I think this is true for any entrepreneur, timing is a big part of anything you do.
0:26:58 I think timelines are especially warped in healthcare because it just takes a long time
0:27:00 to adopt new technologies.
0:27:04 There actually is a peer-reviewed study of the average number of years it takes for
0:27:08 new technologies that are introduced into the medical setting to become mass-market
0:27:09 adopted.
0:27:10 Oh, how fascinating.
0:27:11 Wait, wait, let’s guess.
0:27:12 Two years.
0:27:13 17 years.
0:27:14 No!
0:27:15 I mean, we still have fax machines.
0:27:16 That’s true.
0:27:17 We still have fax machines.
0:27:18 We still use the same…
0:27:21 But we’re not talking about when technology leaves, but you’re right.
0:27:22 It’s the same thing, really.
0:27:23 Yeah, so it gets replaced.
0:27:26 Yeah, you can think about it as all the things that have tried to replace the fax machine
0:27:28 or not yet mass-market adopted.
0:27:29 And it’s the same.
0:27:30 You could see it in…
0:27:35 I think the study actually focused primarily on stethoscopes and thermometers and things
0:27:38 that literally have not been redesigned for hundreds of years because it’s been so hard
0:27:39 to disrupt them.
0:27:42 Over the last 17 years, there’s been a bajillion better versions of the stethoscope that we
0:27:43 are just not seeing.
0:27:45 The wheel could have been reinvented, but better.
0:27:46 Absolutely.
0:27:50 Those are the tangible examples, but the same applies to software and technology.
0:27:54 And that’s a lot of the reason why you see the market-leading companies that own the EHR
0:27:57 space today are literally 45 years old.
0:28:01 And by the way, those companies also didn’t hit their stride until like 20 years into
0:28:02 their journeys.
0:28:06 So time functions completely differently, basically, in this system.
0:28:07 It’s almost like…
0:28:08 It’s like a wormhole.
0:28:12 And second of all, it’s an incredible testament to the strength of these systems that…
0:28:13 Totally.
0:28:14 Yeah.
0:28:16 It’s like, once you do make it, it’s totally sticky.
0:28:21 The LTV, essentially, of tech companies that actually make it and get to a certain level
0:28:23 of scale is through the roof.
0:28:25 There’s no incentive to rip them out because if they work, they work.
0:28:29 The switching costs because of all the human and cultural elements that we described is
0:28:30 huge.
0:28:31 Yeah.
0:28:34 So the longevity of your company, if you’re looking at success, is also incredibly promising.
0:28:35 Yeah.
0:28:38 I mean, certainly at Kairis, the way we mitigated it was we thought about what our fundraising
0:28:42 strategy would be to give ourselves enough runway to have that model play out.
0:28:47 We needed to fund the sales cycles and the adoption cycles to create a new category of
0:28:48 solution that didn’t exist.
0:28:51 Did it hang out in the wormhole for a while?
0:28:52 It’s a big oxygen tank.
0:28:53 Yes.
0:28:57 Global happens in healthcare in under three years, and so you kind of have to give it
0:28:58 some runways.
0:29:01 This is one of the things that we’ve spent time talking about is what does a minimal viable
0:29:03 product in healthcare look like?
0:29:04 Doesn’t exist.
0:29:05 Big bang.
0:29:09 You’ve got to go in and you’ve got to create a category and you’ve got to get that adopted.
0:29:14 I think in other industries, you can sort of quote-unquote get away with having a product
0:29:18 that does one thing really, really well and then start there and yes, expand over time,
0:29:23 but at least you can get by and to prove your value with that initial use case.
0:29:26 I think going back to a lot of the points you made earlier in healthcare, when you’re
0:29:32 in the flow of impacting a patient encounter and saying you’re going to rip something out
0:29:36 or change the way that you’re doing something or what have you, you have to make sure that
0:29:39 it’s going to give you the right answer, so to speak.
0:29:43 Even if it’s just one feature, it might mean, okay, yes, it could be one feature, but you
0:29:46 have to be integrated into seven different systems to make sure that the data flowing
0:29:50 into that one feature is enough to inform the right outcome or decision.
0:29:52 So really fully baked.
0:29:55 If a transaction falls through the cracks, while you’re doing some kind of revenue cycle
0:29:59 type encounter, you might not get paid for a procedure that could have a severe impact
0:30:01 on your bottom line.
0:30:02 You need more funding.
0:30:04 You need to think differently about your strategy for product and what that footprint
0:30:05 looks like.
0:30:08 You have to have the full solution.
0:30:12 And the related point I would make to that is it’s really hard to have a point solution.
0:30:14 Even if that point solution is very, very good.
0:30:18 I think people in general in the healthcare system are looking to buy a complete solution.
0:30:23 So if you take the problem from A to B to C to D, that’s great, but they need A to Z.
0:30:25 They can’t get A to Z from you.
0:30:27 It’s very hard to get them to buy A to C from you.
0:30:29 I’ll go even further than Julie.
0:30:32 I will say, not only does MVP not exist in healthcare, I would argue that product-market
0:30:34 fit doesn’t exist in healthcare.
0:30:35 What do you mean by that?
0:30:40 The definition of product-market fit is when the right product meets a good market.
0:30:44 All of the things we talked about creates such distortions in the marketplace that by the
0:30:49 time you actually get through all the hoops, you have such a skewed product.
0:30:51 It’s not really product-market fit.
0:30:53 It’s almost like accepted product capture.
0:30:56 Here you have regulatory issues.
0:30:57 You have pricing concerns.
0:30:58 You have incumbents.
0:31:03 You have so many aspects that distort the market that I would argue that you don’t have a normally
0:31:06 functioning market for software in healthcare.
0:31:12 How would you both embrace that distortion early on and not get completely knocked off
0:31:14 your path by it?
0:31:19 It strikes me that a lot of what you’re describing is know-thyself, know-yourself very deeply.
0:31:22 That was the tagline I know, by the way.
0:31:25 Oh, was it really?
0:31:26 That’s really funny.
0:31:29 I did not work that in for you.
0:31:34 But also know where you’re going and do that deep, I want to say, soul-searching on a company
0:31:37 level and build out accordingly.
0:31:40 How do you get that big center of gravity of really knowing yourself, knowing where
0:31:44 you’re going, but be able to be flexible with that distortion along the way?
0:31:48 The only North Star you can have, and this is going to sound cliche, but really understanding
0:31:54 your value proposition truly from the customer standpoint, it becomes a critical guide for
0:31:55 what you do.
0:31:58 This is a debate that healthcare companies have all the time, which is should your value
0:32:02 proposition be I’m going to save the system money because the healthcare system is very
0:32:05 inefficient and it runs on very low margins generally.
0:32:08 Should it be that I am going to result in better outcomes for patients?
0:32:13 Is it going to be I’m going to create some sort of a lift in terms of return on investment?
0:32:16 There’s a bunch of different ways you can think about value proposition.
0:32:20 If you don’t have that crystal clear from the outset, the amount of obstacles that you
0:32:24 are going to hit along the way are going to make it such that it’s going to be very difficult
0:32:25 to get to the other side.
0:32:29 If you don’t really understand the workflow and the culture and the regulation and the
0:32:33 governance and the politics and all of the other things, you can have a theory on what
0:32:37 the value proposition is, but you need your customer to confirm that early on and sadly
0:32:40 the best way to confirm that is to have them buy something, obviously.
0:32:44 Julie and I have had this debate before, which is a lot of the software platforms that go
0:32:49 into healthcare have been sort of predicated on we’re going to cut costs.
0:32:54 I don’t know of any sort of solution out there that has meaningfully been able to make a
0:32:56 very, very strong case that they can cut costs.
0:32:59 And by the way, part of it is, I think, is because it’s really hard to measure costs.
0:33:03 It’s almost like a necessary evil where you have to say in some way, shape, or form you
0:33:06 are going to reduce costs, but that can’t be your primary value proposition.
0:33:09 Because at the end of the day, it’s aligned in the cost structure that can get wiped out
0:33:13 over time and potentially get commoditized.
0:33:17 So is the takeaway, know your value proposition as early as possible and test it?
0:33:20 That and then have the conversation of like, okay, if we are able to accomplish what we
0:33:23 just described, is it worth it?
0:33:24 Is the juice worth the squeeze?
0:33:29 Because it’s so expensive to distribute product in this market because of the sales cycles
0:33:34 and the nature of the enterprise sales motion and whatnot, that if you’re not able to envision
0:33:39 a path towards being like at least a half a million dollar kind of a year type solution
0:33:42 in this space, it’s actually not financially worth it to build a business in that area.
0:33:43 Right.
0:33:45 Which goes back to your point of like run the numbers, basically.
0:33:48 At least like back at the envelope, like, you know, whiteboard kind of thing.
0:33:52 I mean, is there anything that you can figure out as you go?
0:33:57 It sounds like you need to know so much before you begin and be so self aware and so kind
0:33:59 of like have the end game in sight.
0:34:03 Are there things that you can leave sort of more organic and like feel out as you go?
0:34:04 Yeah.
0:34:05 No, I mean, absolutely.
0:34:08 There’s tons of things you can be doing on a daily basis with end users and just like
0:34:12 feedback mechanisms on like how people are, are they actually able to do their jobs, for
0:34:15 instance, and making minor tweaks to the workflows and whatnot.
0:34:19 So that was always, you know, a component of a more organic and dynamic aspect of how
0:34:20 we did things.
0:34:24 The other thing that you need to kind of think about doing in parallel is, you know, so much
0:34:29 of success of technology and healthcare is predicated on integrating into other ecosystem
0:34:30 players.
0:34:33 And so this is actually probably one industry where you definitely can’t like just build
0:34:34 in a vacuum.
0:34:38 You actually should understand, even if it’s not, you know, for another few years that you’re
0:34:42 really going to have to do this, like who are the players, we just need to get to know
0:34:46 so that we’re on their radar when time comes for us to take the hammer and like try to break
0:34:50 down the wall of integration with that vendor that we are on their good side and that they
0:34:53 know who we are so we can kind of make that happen faster.
0:34:58 So things like that, I think you can be doing in parallel to, you know, the kind of formulation
0:35:00 of what the footprint of the product is.
0:35:04 If you’ve got the right solution, you can get very creative in how you get paid.
0:35:09 So figuring out different pricing structures or value capture mechanisms, I think is something
0:35:13 that you can do pretty organically because if you are making a difference in the system,
0:35:17 the system has so much cost built into it and so much revenue flowing through it that
0:35:19 there are ways to be very imaginative there.
0:35:21 So that’s the first thing I would say.
0:35:26 The second thing I would say is thinking about adjacencies, you know, going from one, you
0:35:32 know, your core function to the next adjacent use case, not all adjacencies are created
0:35:33 equal.
0:35:34 One might be easier than the other.
0:35:38 It’s almost like, you know, jumping on stones across a pond or something, right?
0:35:41 What’s the next stone I can jump on that’s least likely to make me fall into the water,
0:35:42 right?
0:35:43 Yeah.
0:35:44 Even if it doesn’t get me as far as another one.
0:35:45 Right.
0:35:46 Always have that closer spot insight.
0:35:49 You’re almost, you’re creating the next thing and the next thing and the next thing and
0:35:50 you build out from there.
0:35:54 And eventually you cover so much surface area that, you know, you become a very sticky solution
0:35:58 and you hopefully become a complete solution sort of closer to the A to Z type vision.
0:35:59 Okay.
0:36:00 Last question.
0:36:01 Biggest takeaways.
0:36:04 Quick lightning round for your founder struggling right now.
0:36:05 What would you say?
0:36:06 Bullet points.
0:36:07 Know your market segment.
0:36:12 Be very specific about what segment you’re going after because that has major implications
0:36:14 for your go to market and your product.
0:36:15 Good one.
0:36:16 All right.
0:36:17 Let me get that to you.
0:36:20 One is build the multidisciplinary team early.
0:36:24 Two is understanding and if the person that suffers from the pain point can actually pay
0:36:29 for your solution because there’s a lot of misincentives in the healthcare system.
0:36:34 And three, with the right technology, you can have massive impact on patient lives and
0:36:38 the experience that we have with the healthcare system, which we will all touch in our lifetime.
0:36:42 And if there’s anything you can do to make it better as an entrepreneur, I would say
0:36:44 that is extraordinarily satisfying.
0:36:45 That’s fantastic.
0:36:46 And good bullets.
0:36:49 Thank you both so much for joining us on the A16Z podcast.
0:36:50 Thank you.
with Jorge Conde (@JorgeCondeBio), Julie Yoo (@julesyoo), and Hanne Tidnam (@omnivorousread)
Building a software company in healthcare is hard — and comes along with unique challenges no other entrepreneurs face. In this conversation, a16z bio general partner — and previous founder of genomics company Knome — Jorge Conde; and a16z bio partner and former founder Julie Yoo (of patient provider matching system, Kyruus) share their mistakes and hard earned lessons learned with a16z partner Hanne Tidnam.
Why is this so damn hard? How should founders think about this space differently? What are the specific things that healthcare founders can do — when, where, and why? You’ll wish you only knew this when you started your own company!