Commure Hits $7B Backed by GC, Sequoia, Morgan Stanley

[00:00] We've raised $70 million on a $7 billion valuation led by General Catalyst with support from Sequoia and Morgan Stanley. In total, the business has raised about $750 million. We support 200 million patient encounters every year now, saving at least 75 million hours a year for physicians. The progress over the last 18 months in the business has been rapid. We've expanded in geographies, we've expanded logos, we've added some of the largest health systems in the country. [00:30] agents to back office automation. In health systems like HCA, we've been able to add a couple hours to every physician's day. And this is a company that generates north of $100 billion in top line revenue. LLMs are this gift that we've received to go nuke all of that work tax. And that in some ways is going to be my life's work in terms of just eliminating the work tax and making the system a trillion dollars more efficient. Because if you do that, you make the entire American middle class well-efficient. [01:00] you [01:03] Danai, welcome to Sorcery. Thank you for having me. Thank you for having us here at one of your investors' offices. Where are we? We are at the Human Capital office. Their Human Capital is run by two of my closest friends, Barsh and Arman, and they really helped start Comure and have been with me on this journey for [01:18] Six years now. [01:19] And it's beautiful. It is beautiful. It's a beautiful journey. It's way prettier than our, you know, dingy Mountain View office. So we use this one instead. Okay, we have a big congratulations for you today. [01:29] You've raised $70 million on a $7 billion valuation led by General Catalyst with support from Sequoia and Morgan Stanley.

[01:37] What happened? [01:39] We raised around, thank you. And, you know, I think the progress over the last 18 months in the business has been rapid. We've expanded in geographies, we've expanded logos, we've added some of the largest health systems in the country, and we've helped physicians save time. [01:53] and physicians make more revenue every single day. [01:56] Based on that growth, we raised at a pretty large premium to the last price and have [02:02] you know, now really going to use those dollars to accelerate R&D, accelerate some of our investments in further model development for everything from voice agents to back office automation. It's a massive trillion dollar problem and you can build for decades. You started the company at 18. You had a bunch of M&A. [02:19] in the process of that, and how many [02:23] How many employees do you have now? [02:24] The full company today is about 1,200 people. Oh my gosh. Yeah, we're seven different offices, India, Bangladesh included. [02:32] And then a New York hub, a Boston hub, a Nashville hub, and then Mountain News HQ and Salt Lake as well. [02:39] And you raised how much in total? In total, the business has raised about $750 million. Mm-hmm. [02:45] Okay, well, what makes this round different than other rounds? [02:50] I think, you know, this round one, we really didn't need the capital. We raised it. [02:55] for pricing purposes and we saw a couple interesting opportunities to accelerate R&D, particularly around AIR, which is our LLM native EMR platform, and our voice agent platform. [03:06] And so the idea is if we can accelerate timeline, bring in a team of 40, 50 killer engineers,

[03:12] to attack some of these problems that are adjacent, but converging upon the same solution set that we have in RCM and Ambient, it's going to be net better for our customers. Number two, we've extensively used CVF, which is General Catalyst's non-dilutive customer value fund, to fund our go-to-market expansion without needing to dilute shareholders. And so the last two years have been very non-dilutive for the company and for shareholders. And this was, you know, a pretty non-dilutive round as well against the valuation. We raised not a ton of capital because we didn't really need it. [03:42] it lets us sprint and continue building. [03:45] I want to talk about CVF. We had Hemant on not too long ago, and we talked about how General Catalyst is not a venture fund. They're a company. They're a huge platform. And I also just had Teresa Carlson on, and we talked about General Catalyst Institute. She gave you a nod, too. [04:15] how you got introduced to it, what was the sell there? - Yeah, I mean, I think if you, [04:19] back up the timeline super aggressively. And you go back to the father of capitalism in some way, Rockefeller. Rockefeller was the OG when it came to using credit. And he really didn't dilute himself or his shareholders much at all. [04:31] And they aggressively knew that, hey, if we drill in these places and we drill for 10 years and we develop our refineries and we develop our supply chain, these many dollars are going to show up. And we're going to use credit to fund that expansion as opposed to taking on more and more dilutive equity. And CVF is really, I think, built along that concept. And credit is not a unique concept in a lot of ways. The reason startups shy away from it is it often puts this massive liability on your balance sheet.

[05:01] - Yeah. [05:02] you're now, your whole company could implode. And CVF is built to address that need very particularly, which is, [05:09] Instead of risking against balance sheet, it's risk against forward-looking customer cohorts and performance. So if you're confident that your SaaS cohorts will perform and revenue will show up, and when you put in go-to-market dollars, net new logos will show up, you can use, you can essentially forward pull that revenue. [05:25] to fund expansion faster and faster. [05:28] And I think this is the purest form of go to market. You actually don't want to use... [05:32] balance sheet for go to market. You want to use balance sheet for R&D and these more durable long-term investments. [05:37] And that's why we used it and I think that's why it's an amazing vehicle. [05:41] Wow. [05:41] Yeah, I haven't talked to many people about that. We do a lot [05:47] of equity fundings and then we've done some public companies but there's definitely more [05:53] nuances to the types of capital you can buy now, especially as tokens increase costs within companies. Are you seeing that? Are you guys using lots of tokens? Is that compressing your margins? What are you? I think it's a great question. And there are a lot of businesses where [06:08] You know, you look at [06:09] VCs have a good sense of what a SaaS business looks and feels like in terms of its margin structure, its payback, its CAC, its LTV. And one of the challenges is that these token businesses have essentially turned a lot of software businesses to service level margin. And if you play the game the same way you did two years ago, [06:25] you can blow up and you can burn a lot of cash in the process. [06:28] And I think good operators and good CFOs have a fundamental understanding of their margin structure and their payback structure.

[06:35] When we use CVF, we have parts of our business that run at pure SaaS margins in the high 80s, mid 90s. [06:42] And you have other parts of the business, like Full Cycle RCM, [06:44] that have a labor component and a heavy token component. And those run at anywhere from the [06:50] high 60s in their implementation stage to the mid 70s in their steady state. [06:54] That's a different margin structure than SaaS. It's still great. You can, you know, it's still, you know, many businesses would take 70% margins like with every day of the week with their eyes closed. But I think the, you know, for us, the question is, [07:06] can you, if you fundamentally understand the payback periods, then you should use CVF. You should use non-dilutive mechanisms to fund that expansion. [07:13] You mentioned before that you want to go public. Is that still the case? 100%. I think the best American businesses go public and become long, durable parts of [07:22] retail and allow the public to invest in them. I also think [07:26] Given the size and shape of our business, we're in the hundreds of millions of dollars in revenue. We support some of the largest systems in the country. [07:33] Our acceleration in some way would be fueled by going public as opposed to being hampered by it. [07:39] Kabir is building the AI powered OS for healthcare. Can you unpack that a little bit more and the state of healthcare? [07:47] Yeah, I think that's a great question. So [07:49] for Cumura at its core, [07:51] is a software company that builds tools for providers and for healthcare administrators. This is a multi-trillion dollar problem and even when you cohort it down, there's a trillion dollars spent on healthcare admin every year. That's clinical admin, that's

[08:06] financial admin, like revenue cycle management, all of the tasks that happen in the back office of a health system, like scheduling prior auth appeals. We've built a series of fine tuned language models. [08:18] and agents that automate all of those tasks and truly take a health system that might operate at [08:24] 2% or 3% operating margin today because of all the labor needs. [08:27] and turn it into something that can operate at 20-30% operating margin, like the top percentile healthcare practices and systems. And our belief is that that's not going to be 100 different point solutions. That's actually going to be one revenue engine and a series of agents that are orchestrated on a single platform, which is ComirOS. [08:57] their revenue cycle and all of their back office needs. [09:00] and see that boost in margin because they can grow without needing to have labor directly track their top line, which is [09:06] Historically, when a healthcare practice grows, you just linearly tack on more and more labor. [09:10] In health systems like HCA, it's a very similar story. We've been able to add a couple hours to every physician's day. And this is a company that generates north of $100 billion in top line revenue. So if you can... [09:23] add even a couple percentage points of juice to that, that is billions of dollars of operating cash flow that get created across the country and obviously across HCA's empire as well. I think the state of healthcare AI today is really interesting because

[09:37] You have the early adopters, the physicians, the practice owners, the health systems. [09:42] that use 2025 to really leapfrog ahead. [09:45] and turn into these bastions of what an agentic healthcare system looks like. And for certain parts of the systems, [09:51] it really feels like the future. Documentation is automatic. Billing is automatic. Coding is automatic. All these tasks are now, you know, entirely done by software. Other systems, the vast majority of the country, if you look at rural health care, if you look at a lot of the nonprofits, a lot of the academic institutions, [10:08] have been slower to adopt. I would actually say that healthcare's biggest problem today is [10:14] It is run by politicians. It is not run by ruthless, pragmatic business operators. And if you look at the for-profit institutions that are growing the fastest, that are compounding the fastest for shareholders and for patients, [10:27] The single trend that's shared amongst all of them is you have amazing, aggressive CEOs at the top, as opposed to design by committee and decide by committee, which a lot of your... [10:37] you know, smaller academic nonprofit systems historically have used. So for me, the most important thing in the next five, ten years of American health care, if we're truly going to build a durable system, is [10:48] We need a shift into all operators of health care systems and hospitals need to shift their mindset to being business managers and being very aggressive CEOs. [10:57] as opposed to being local politicians, which has really been the game for the last 20 years. And you're at over 100 million in patients annually. We well north of that, we probably support

[11:08] I want to say 200 million patient encounters every year now. Oh wow. And the, [11:13] for just patient engagement, Camero Engage, which is one of our, it's a language model that can speak to the patient, help schedule the appointment, reschedule, cancel. We do about 100 million patient interactions on that one product every year. [11:25] ambient documentation, which essentially [11:28] You press a button, it listens to the appointment. [11:30] A language model summarizes the appointment. [11:33] plugs it into the EMR, and then generates the super bill, which you send to insurance. That does about 50 million annualized appointments every year. [11:40] saving like, I want to say at least 75 million hours a year for physicians. And that's just productivity straight back into the American economy. How did you scale that up so aggressively? Do you have specific partners? [11:52] unlock more volume? Like, how does that work? [11:55] there's really [11:57] two or three components of that business. One is the PLG motion, which is a physician literally finds the product online, loves it so much, starts using it, tells the other physicians in their practice, and then we go sign an enterprise agreement and expand organically within the system. And then there's a more top-down approach. We have a deep relationship with both HCA and Tenet, who are two of the largest for-profits in the country and two of the largest businesses in the country. And there they own a lot of hospitals, they own a lot of practices. HCA alone owns about [12:27] of ambulatory care. And we've worked with the management there to essentially propagate these tools [12:32] through the EMRs and through the software across [12:34] all of their whole empire. [12:36] And then finally is EMR partnerships. So we have a deep relationship with MetaTech as well as a really, really close relationship with Epic. We're a toolbox partner and at the press of a button, an Epic physician can turn on Khmira Ambient, start using it.

[12:51] They can also turn on autonomous coding and start using that for the next piece of the revenue cycle. And I think because we've partnered with the EMRs as opposed to just entirely try to disrupt them, we've had a really fruitful relationship. [13:03] in expanding through that as well. [13:05] because you're going after a unified platform versus a point solution, you're increasing the complexities by like, [13:12] For sure. [13:13] Crazy multiples. Yes. So with each one of these verticals that you start, how do you [13:18] drum up those teams and then condense the stack. [13:22] I think this is [13:23] It's one of the big debates, honestly, in our company, which is, uh, [13:27] you the beauty of a platform is you can deliver definitive ROI across the whole continuum of care. And there's no question [13:37] that when you deploy the full Kumira Thales platform, you get 15, 20% more revenue at the other end of it because everything is connected. It's a way more complex implementation, though, and it's a way longer implementation. And so our strategy has been twofold, which is. [13:51] One, have a series of point solutions like ambient documentation, autonomous coding, intake. [13:57] that can be those wedge products that enter a health system or enter a practice. [14:00] and allow for quick adoption. [14:02] And then the platform sell, which quickly follows, but might be a 12-month sales cycle or longer, a 24-month sales cycle for the larger health systems. [14:11] But, [14:12] contracts that are worth tens of millions of dollars at the tail end of it. So getting good at doing both, because there are companies where all they do is [14:20] that one point solution and it's a three month sales cycle and it can grow quite quickly.

[14:24] we've had to be very disciplined about [14:27] ensuring that we focus go-to-market resources on both [14:30] And that these two motions support one another as opposed to, you know, being just two completely different parts of the business. [14:36] Why are you doing that? I think the long-term bet is [14:41] the final ACV capture in the space is going to be the company that provides that full platform. Here's a great example. When we launch our platform in a 100 provider practice, the first thing we do is we displace the point solution scribe that they're using. And so that company's revenue just went to zero. And our revenue just went up by a factor of 10 on the account. So [15:01] If you look at where the world heads, these things do tend to consolidate. Whenever there's a paradigm shift like language models, [15:08] you will have eventual consolidation on the one platform that provides them all the tools in a unified manner. And yes, you could probably get ephemeral revenue right now. There's a lot of point solutions garnering a lot of interest. There's a lot of nonsense GPT wrapper businesses out there right now. [15:23] say them by name, maybe I'll say them by name later, but they grew fast, they got a lot of VC interest, [15:29] but they're going to be gone in three years because companies like ourselves [15:32] will completely disintermediate them. [15:34] And this happened before, you know, in the 80s and 90s, there was a software that came out called Grammatic, and it was a auto-correcting tool. And every, you know, all the VCs of the time loved it because it was growing so fast. You plugged it on top of, you know, the word processors on the early computers, and everybody loved it. [15:51] And then Microsoft Word came out with AutoCorrect as part of the software, and this company's revenue went to zero.

[15:58] And this happened again and again and again in multiple verticals. And the VCs just kind of forget. But if you study history and you're a student of business, I think you see that [16:07] these point solution businesses, [16:09] either have to turn into platforms very quickly or they die. [16:12] I was just going to ask you before this amazing question, what's going to happen to point solutions? Yeah. That. They're going to die. We will. In health care, I think we will kill most of them. And it is we do it with a passion. Like it is something that we care about deeply in our business, that these multiple point solutions that create [16:35] chaos for the IT department and for the healthcare provider [16:38] need to be eviscerated and turned into a consolidated platform. [16:41] because of [16:43] the regulatory environment that healthcare sits in, and this is a Teresa Carlson question, but what does the actual process of AI implementation look like in healthcare today? And what does that look like transformationally in the next five years? Because it's a different point in which, you know, we were joking earlier, I was like, [17:03] today, congrats on your consumer company, blah, blah, blah. But like, it's totally different than like a consumer company, like an AI thing, but like, [17:11] deep-seated regulatory environment, what does it actually take to get this off the ground and where are we? Yeah, first of all, Teresa's a legend and she's a board member. And I said this before, but nobody shoots whiskey like Teresa. The woman is truly a legend in many ways. And I think the, you know, when you listen to her stories,

[17:30] from Amazon, which [17:32] they did one of the hardest things in the world. They went to the CIA, and they went to the governments, [17:37] And they told them, you guys are going to turn into [17:39] cloud-first, cloud-native organizations. This is insane. Imagine going to an IT team that literally has file cabinets and scanners and massive on-prem mainframes and telling them, no, [17:53] the company that you buy your books from is actually going to be the company that transforms you and turns you into a digital native organization. And she did it in partnership with Jassy and the whole team there. [18:04] And, you know, I think the learnings, whenever I've talked to Teresa about this, the learnings have always come down to, [18:11] You've got to align it. It comes down to dollars and it comes down to ROI. [18:14] You have to be able to sit in a room and you have to tell a very clear story of what the world turns into three years from now. And it's investment. Like when we sit in a room with HCA, [18:23] they're investing millions of dollars in resources, in implementation time, in training time, in physician overhaul, to make sure that people use these AI tools correctly and they turn into an AI native organization. And that's another reason why I think many of these point solutions will fail, even if you get some early adopter bottoms up usage. [18:41] you'll never get that organizational change. There's no version of the world where AWS could have been a PLG cell. It needed to, it really needed that enterprise motion where [18:51] Theresa and Jassy and Garmin and the team at the top came in and sat down with these government officials [18:57] and pitch them on the vision of digital transformation. And now we're doing the same with language models. And the best teams are the ones that know how to tell that story,

[19:06] paint a very precise picture of transformation, [19:09] bring the IT teams along rather than leave them behind. Because this is such a sensitive environment with very sensitive data, life or death information, it's critical. So where does accuracy and reliability sit in decision making? Yeah, I think the [19:27] where you see rollouts get completely stopped or companies go to zero overnight, [19:32] is in cases where a model or a tool [19:35] hallucinates or performs incorrectly and ends up impacting patient care. [19:40] And so it ends up being maybe the most important thing. [19:43] And the only way to truly ensure that is one, great evals. You need to have a massive data set of [19:50] historical documentation or whatever the task might be, coding, voice calls, [19:56] to then measure your model iterations on. Because when you release a new version of a model or a new version of an agent, [20:02] It does some things amazing, but then it does a couple of things that are kind of weird. And there might be regression and performance deterioration. [20:08] on the fringes and in categories that you weren't even thinking about. [20:11] And I think in smaller companies or in companies that don't have a lot of engineers, you can ship quick and get this stuff out. [20:17] But that performance degradation goes unseen and then turns into just cascading problems for organization later. So our belief is we work hand in hand, one book with the regulators, but I think more importantly with the customers, to show them that the edge cases they care about always work in the test harnesses, always work in their backtesting data, and then make that super public. And then at the same time, like when there are inevitably hiccups, you need to have a very fast way to iterate

[20:44] and give feedback to the model and patch it so that you're not sitting around with a, you know, faulty [20:49] you know, hallucinating agent in a healthcare setting. [21:19] treasury and FDIC protection into one powerful account. You can send and receive money globally at lightning speeds, get 20 times the standard FDIC coverage through their partner banks, and even high yield from day one. With same day and even same hour liquidity, access your funds anytime. Companies like Scale AI, DoorDash, Service Titan, Hims, Anthropic, [21:49] That's B-R-E-X dot com slash sorcery.

[22:19] dot com slash S O U R C E R Y. [22:22] I know AI insurance sounded really silly when it was just first rolled out, but it's clearly deep-seated in lots of [22:30] companies. Is that something that you have thought about? How do you think about that? You know, I think [22:39] my [22:41] prediction for the space is it's eventually all going to converge upon some sort of like [22:46] general liability software cyber insurance package. And the reason is, is that [22:51] I think in most cases, the language model actually-- widespread use of the language model [22:56] in healthcare in particular, actually improves outcomes enough, where the malpractice exposure, I mean, it's just going to be a better look for, it's going to be a better end outcome on a dollar basis too for the CFO department and the CLO's department. [23:10] When we talk to Health System C-Suite, [23:13] The CFOs and the CLOs are so excited because they see in the data, just like with self-driving, that yes, occasionally you have weird performance. [23:21] But on the averages, and like in the 99% of cases, it improves outcomes so much. [23:27] that the malpractice implications are actually very net positive. [23:31] And I think because of that, you should see a reduction in premiums. And maybe there's like an AI subcategory that emerges in general liability or cyber insurance. I don't think it's going to be this... [23:41] category of its own. Because in terms of pricing it, it's no better or worse than general software for most industries. Not to go too far off the line, but what is the implementation process with these teams? Do they have the technical rigor to understand? Obviously, AI, this whole thing is very new. It's not just you're deploying software, but you do need to be trained for it. So how do you do that for the different partners that you work with?

[24:11] So we have this forward deployed team of engineers that, you know, obviously stolen from Palantir, but I mean, the model works so well. You take... [24:19] smart, hungry people that are early in their careers, and you throw them into the face of the problem within the hospital, within the health system, [24:26] um and you you know you you kind of let them free and and that has been [24:33] one of the ways that we've actually propagated usage of these models very quickly, because a physician sees this [24:39] excited young engineer come into the hospital and wants to co-develop with them and wants to work with them and wants to iterate on these models. And so our first implementations with HCA was literally a couple engineers on the ground in a hospital with three physicians. And it was like two months of back and forth. How do we make this model work really well in a hospital setting or an ED setting, which are really [24:59] hard settings to perfect documentation in. And [25:03] Once it started working super well, we expanded it to another group and then another group. And then there was a top-down decision to go enterprise-wide. And the implementation teams [25:13] Now it looks a lot more like just general software trainers. They come in, [25:16] They help launch, they teach them how to speak into the microphone, what to say, [25:21] just really be natural in your conversation with the patient. And then on revenue cycle, we'll usually have a team that goes into the billing department, [25:30] migrates their system using a set of agents that'll take, you know, old system of record and populate it into new system of record. [25:36] and then really show them how to [25:38] Invoke an agent to do a task for them. The beauty of I think LLMs and agents is it's quite intuitive. Everyone

[25:44] Anyone who's worked in a corporate setting for years and years has a sense of how to delegate and how to give someone a task. You pass them context and you give them a direction. And I think in healthcare, [25:55] it's even simpler. It's usually the context is [25:59] the claim and the encounter history and the patient's chart information. And then the direction is whatever you want, you know, the person to do, which is resubmit the appeal, work on the denial, file the prior authorization. [26:12] And we've been able to get these models to get very, very, very good at these tasks because of the sheer volume of data that we have. And so the training ends up being maybe a day, day and a half in person, and then [26:23] you see people just run with it and eventually just train each other as well. [26:26] Have you seen the composition of healthcare teams change at all? I mean, I know this is probably still too early, but it is really, it should be improving their work, [26:35] load a lot because it's really redundant, it's handwritten, but how are you seeing compositions change and what do you think it'll look like? Yeah, I think to the first point, [26:45] The feedback from the physicians and from the providers and the medical billers is insane. We had a quote last week that Heymont actually posted, and it was from a physician that said, "I was about to quit. [26:58] I was looking for another job because these 14-hour shifts were just [27:02] ridiculous and it was getting, you know, I couldn't take care of my kids and it had just turned into too much for me. [27:07] And then I used the tool, the Cremere Ambient tool, paired with RevenueCycle, [27:12] and it literally finished 14 charts instantly.

[27:15] And she was like, boom, I have two more hours in my day. I can do this now. And decided to stick with the job. Like literally didn't leave health care. We have a massive labor shortage in health care. We cannot have trained people leave the system. And this woman did not leave health care because of the tool and because of a language model and because of our software products. So that's been amazing to see. And you hear stories like that every day where... [27:37] nurses, physicians, medical administrators feel like they have another 10, 15 years they can give back to the community. [27:44] because the tools have given them that. [27:45] Number two, in terms of composition, I think the smartest health systems [27:49] are starting to hire these [27:51] AI implementation teams are starting to hire engineers of their own to come in and help with implementations. [27:56] Now, the only systems that can do that are the ones that, it's kind of like a chicken and egg problem. The only systems that can do that are the ones that, [28:03] generate enough operating cash flow to forward invest. [28:06] And that's sort of our bet is step one. [28:09] give them the tools that improve cash flow, and then that cash flow lets them invest in R&D teams, [28:15] who will then go build more and more interesting stuff in partnership with us within the hospital and within the system. SUMA is a great example of this in partnership with GC. What is it like working with SUMA? It's a real hospital? It's a real hospital. It's a real hospital in Akron, Ohio. And, you know, I got to hand it to Heman. The... [28:33] The beauty of GC and I think Hémant is [28:36] You're so ambitious and they are not content being a VC firm that makes just great returns for LPs by investing in a series A and, you know, the two or three companies that matter every year getting into the rounds. They really want to.

[28:49] Bend the universe. And part of that means going to Akron, Ohio, buying a system that was [28:55] on track to go bankrupt in the next couple of years and transforming it with AI. [29:00] There are a lot of people that would be more than happy to sell tools to the miners and make a lot of money doing that. I think it takes a special person and a special organization to say, no, we're actually going to go direct. [29:10] and we're going to show the world that you can build a better hospital, and you can build a better health system, and you can do it today. You don't have to wait around and just make your money on software. We had RFK Jr. visit the hospital last week, and we hosted him there, and it was an amazing visit, showing him the transformation with ambient documentation on ComirStack, with ADOC, which is an imaging-based company that's, [29:32] essentially automating a lot of the scan process, [29:34] And so one of the things that we've noticed at SUMA is, [29:39] the appetite from the physicians is just, [29:42] I mean, they're ready to go. They want those hours back in their day. They want those dollars to show up in the system to create that cash flow cycle. [29:49] And I mean, it's been so much fun on the ground. We have 10, 15 engineers there at any given point of time. And we're working hand in hand with the physicians. This might be a total tangent, but [30:00] I travel a lot, so I'm often fatigued and like over exhausted to the point where I like all pass out. And because of that, I get IVs to my house. Nice. And I always chat them up. It's always really fun because they have the craziest stories, these nurses. And most of them are ER nurses. And this woman who came to my apartment like a couple weeks ago, I was asking her about her job. And like she was an ER nurse.

[30:30] or do you like coming to, you know, [30:32] nice, calm environments? Or do you like being an ER nurse and like dealing with all the complexities and like the chaos and all of the adrenaline and all that stuff? Yeah. She was like, [30:42] No, I actually like prefer that. Like I like it. Like she's just like wired to do that. Yes. Which is so cool to hear and like know because I mean from the outside like [30:52] Yeah. That scares me. Yep. Yep. But they love their jobs. They love their jobs. And I think [30:59] There's something to it where, in the ER in particular, we've spent a lot of time building our tools for the ER. It is a chaotic environment. And if you're a triage nurse, you are literally making life or death decisions in front of you. I mean, you pick... [31:12] who goes in first, who goes in second, who gets care right in the lobby. [31:16] And those are not easy decisions. And I think the [31:20] you know, the mental fortitude of these nurses, combined with their appetite to just take on the craziest problems, is super inspiring. And it makes it really easy to build tools for them. [31:31] Our engineers, their favorite visits are the ones where they get to go to the ED and sit with the nurses because they just see so many problems, so much pain that they can come and just sink their teeth into and build tools for. [31:45] For me, that's exhilarating. It's like, what more could you ask for? - Yeah, that's crazy. That's amazing. They have the craziest stories too. I can't repeat it, but like, people in California are effed up. - I believe it. - They do weird stuff. - There's a lot of weird people in the world. - Sometimes it involves animals.

[32:03] Which the state needs to change the regulations on. I don't. Anyways. Got it. So I guess since we're already on the regulatory kind of topic. Yeah. Sorry for the transition. People do weird things with animals is what I learned. It's not good. I'm going to take that with me. [32:24] Speaking of regulatory environment. [32:27] - Is the admin more or less acting with urgency? Like what are you seeing with this new change of hands? [32:36] I think this admin is doing an incredible job and they are [32:40] they're stepping out of the way in the areas where you want innovators to just quickly get on the ground and do good work. [32:47] and they're intervening in the areas where regulatory intervention is needed, like Interop. I think one of the [32:54] biggest problems in healthcare over the last 30 years was [32:57] An EMR could guzzle up all your data and then sit there, and then when you ask for it back, [33:02] quote you some crazy fee to move one piece of data from one system to another. It's insane. Like, genuinely, if for all the shit Salesforce gets as a company, [33:12] If I want to move data from one... [33:14] place in Salesforce to another place in Salesforce, I can do it with my eyes closed. Like there's no, I don't have to ask for permission. If I want to do that in Epic, [33:21] I've got to go sacrifice my first born child to Judy. And then maybe I will be granted a meeting to meet with someone who lets me move data from point A to point B. This is insane. And so data liquidity is a very real problem in health care. And I think the admin is tackling it head on. They're not

[33:37] you know, they're not dealing with any more interop bullshit. And they're funding the Department of Justice to come after folks that are [33:45] engaging in information blocking or violating the Cures Act. And as an innovator, that's welcome change. So I love the work of this administration. What do you think the biggest misconception is working with the government? [33:57] I think, you know, we work with a lot of branches of government, with the FDA, on med device and regulatory and algorithms. We work with CMS on payer policy and, you know, everything from what, when you think about the revenue cycle, what sorts of things should be approved instantly and what sorts of things should require an appeal or prior Roth process. And I think the... [34:20] The biggest misconception is it's [34:22] that they're hard to get in touch with or that they won't take a meeting. I've actually found that [34:26] you can meet with someone in the administration fairly quickly, which is amazing. I mean, this is the, this is not how it always was. You can literally send an email in and probably get a meeting in a couple of weeks. [34:38] air out your concerns, [34:40] and get a very structured, rational response back. And nothing in the government is going to get done tomorrow, but there is a timeline and there's a plan. And I think folks like Chris Klomp, who's come in and is leading a lot of those initiatives, are just a breath of fresh air. And to see innovators in chair in government positions, I think most people don't realize how forward-thinking this admin is. I asked Teresa the same question, and she said similarly, like, start earlier. Start early, yes.

[35:10] sooner they want to listen to you. Definitely wear the right outfit. Yes. [35:15] That was a job. Which we'll come to in a bit. Later in the show. Yes. But she did say that. And then I was at a conference earlier today and one of the heads of HHS said, [35:27] Was there? And he said, yeah, like, come, come talk to us. Yes. [35:30] Like, don't wait. They're willing to engage early. Yeah, just start early. Yep. And there's good people. There's good people in the admin, and it's great to see. Okay, so you have 1,200 employees? Yes. You went viral for something that Alfred Lynn doxed you on. I want to ask about how you felt about that after I... [35:53] recited verbatim, but you went viral for what you want to look for in employees. Specifically, heat-seeking missiles, which is super aggressive, but it's an aggressive mission. [36:08] It is. [36:10] So you said describing this person, [36:14] This person actively seeks out the hairiest, gnarliest problems that customers have or exist generally in the business, i.e. go to market efficiency, and then surgically works to eliminate said problem. They have almost an addiction to seeking out sources of pain and blowing them up. [36:32] He's seeking missile for pain. I think, well, first of all, I was honored that so many people resonated with it because it is something we really look for in our business. And it was an email that I had sent out to our whole company describing

[36:45] what it means to work at Khmer. And you have to be a heat seeking missile for pain. And Alfred forwarded to some of the other Sequoia founders and then tweeted about it. And yeah, it went vile. We got-- I mean, from a hiring standpoint, we got some amazing inbound that day. Really? So I'm very grateful that Alfred posted about it. [37:02] For me, I think it's interesting because if you... [37:05] If you look at value creation, like in many ways, [37:07] pain is the genesis of all capitalism. Like there is problem, and then someone comes up with solution, [37:14] And then there is value creation and value capture in that process. Without pain, there is no capitalism. And as a result, I think we should all pray at the altar of pain. Pain is signal. Pain is signal in a business. Pain is signal outside with customers. It means that there is something broken in the human experience and you should go fix it. And that's what humans are wired for. And... [37:34] I think to be successful in starting companies and building companies and working at fast growing companies, if you do not have an addiction for finding pain and blowing it up, [37:43] you won't survive. And the reason is that without [37:47] that mindset of I need to find pain, not just like let it happen to me and then deal with it, but find pain and blow it up. Your company won't grow as fast as your competitor. Your, you know, your go to market will never be as fast as the next guy over. Your products will never [38:00] evolve as quickly because [38:02] Every product has pain. Every organization has pain. And a heat seeking missile for pain [38:07] He's addicted to finding that and blowing it up. So it is, to me, maybe the most important signal on hiring. [38:13] And it's probably also the most important signal

[38:16] in investing and in finding founders that have that addiction to finding pain. [38:22] You started the company at 18. You're not 18 anymore. We clarified that. Yes. Sorry. It sucks, but I'm not. I tried. Good. So with that, the company has been through a couple of rounds of M&A. You've acquired, you've merged, you've done many things. Was that process part of which changed your perspective on how to build a company? Like that is not an... [38:48] that's not a conventional way to build a company. So like, how did that, what is the strategy there and like the process and the thinking? - I think, you know, one of the interesting things is [38:58] If you look at the history of American businesses, the great American businesses were built on the backs of M&A. Rockefeller was maybe the best acquirer of businesses of all time. I would say that Disney, like Iger's Disney, was one of the most acquisitive and well-structured acquirers of [39:15] IP and talent and technology that we've ever seen in the 2000s. I think there's a ton of other just amazing businesses that are worth hundreds of billions of dollars that people don't even talk about that are amazing acquirers. Dell. Dell is another great example. People don't talk about Dell. There's a [39:30] maybe $100 billion worth of value created in the last 15 years. [39:34] through Michael Dell and Silverlakes partnership. [39:36] And they acquired some very interesting businesses, put them together in interesting ways, unlock distribution, unlock products. And I... [39:45] I think it is a skill that as you grow as a founder, you need to get very good at.

[39:49] In our journey, we've used M&A really for distribution. And in some cases, I think for talent, where there have been [39:56] critical people that are building amazing products in a space and we've brought them in and accelerated the whole company as a result in the kumura thalas merger i think it was [40:04] It was really both of those things. [40:07] Heymont had amazing distribution through his General Catalyst health assurance framework and all the hospitals that are partnered with him. He had an amazing access to capital that has really helped an accelerated commure. [40:20] And then I think on the Athol side, we had [40:23] great engineering talent. We had heat seeking missiles for pain that could go in and solve problems. And you put those two things together and there was a lot of value creation in that process. [40:32] We've also acquired some businesses like Augmetics that was a public company, and it was particularly done for distribution. They were in 40 different health systems and trading at like 1.5x revenue. The public markets weren't valuing this company correctly. And we seized that opportunity, brought it in and used their distribution to very quickly expand our products. [40:52] and also transform their margin structure using LLMs. You know, primarily labor-based business that is now primarily software-based using language models. - What's the biggest takeaway lesson [41:02] or challenge that you've had through those processes. [41:06] I think the number one learning in M&A is there's a lot of people that think M&A is about finding the perfect asset or acquiring the perfect asset and diligencing every, you know, making sure that it's pristine. Pristine assets are usually, you know, there's no price disparity. There's no advantage that you can take of a pristine asset.

[41:25] Most M&A, and I think most good M&A, there are skeletons. There are things that you have to internalize that are not working in the company. [41:32] and be okay with that because you have the other pieces that will make it work. In the case of Augmedics, great distribution, great legacy relationships, [41:40] But the pace of innovation wasn't there. And they didn't have a software engineering team that was iterating every week, or in some cases every day, which is what you need in the age of LLMs. And when we brought that in, I mean, it just it created so much value at HCA, it created so much value at Sutter, it's created so much value at our other these other health system partners. And so but we have to be OK with the fact that. [42:03] it was suboptimal in some ways. And we were going to plug those things out and make a quick change. I'd say number two is, [42:10] you have to be very ruthless about stating [42:13] what the culture of the combined business is. You can't, I think this idea of you acquire companies and like everyone's friends and, you know, we're going to be like your little bit of your culture, a little bit of our culture. [42:23] No, there's always [42:25] a winning acquiring culture that becomes the culture of the combined business. And there are people in the company you're acquiring that heavily opt into that. There's also people that opt out and that's totally fine. You got to just give them that opportunity and then start from fresh. [42:37] How long is that window? I think you have like [42:41] probably have like 90 days to make like a very aggressive, maybe less. I would say like 45 days is like kind of set. Like people are now, they know what they're doing. They're either locked in or they're not. [42:51] And by day 90, if [42:53] you know, that reset hasn't happened, you're going to continue to run the business the way it was running before, which

[42:58] probably isn't a good thing if you had a very genuine acquisition thesis around it. We didn't talk about this at all, but you started off as a blood testing company. Yes. And you were inspired by Elizabeth Holmes. I would not say I was inspired by Elizabeth Holmes. That sounds like an Alfred jab. But we were started around the same time that the whole Theranos saga was happening. That is not untrue. Sorry. It's too good. [43:27] VCX by Fundrise, the public ticker for private tech, allowing investors of all sizes to invest in venture capital. View the portfolio at GetVCX.com. That's GetVCX.com. Some of you may not have heard this yet, but our sponsor Public just launched something called Generated Assets, and it brings AI into investing in a way I've honestly never seen before. Here's how it works. [43:57] or defense tech companies growing revenue over 25% year over year. Publix AI then dispatches a swarm of agents that scan every single U.S. stock, evaluates them, and instantly builds a custom index around your thesis. What really stands out is how clearly it explains why each stock is included. And before you invest, you can even backtest your idea against the S&P 500, so you're making decisions with real context, not just guessing. And beyond generated assets, Publix lets you invest in stocks, bonds, options, crypto, all in one place. [44:27] They'll even give you an uncapped 1% match when you transfer your investments over from another platform. If you want to build a portfolio that actually reflects your thesis, visit public.com/sourcery.

[44:37] Paid for by public investing. Full disclosures in the description. [44:41] Enterprise AI runs on Merge, the AI infra platform for integrations, agent tooling, and model orchestration, so your teams ship product, not plumbing. Mistral, Dropbox, and Drada already trust Merge in production. Start building at merge.dev. [44:57] Founders scale faster on Deal. Set up payroll for any country in minutes. Hire anyone anywhere. Get visas handled fast and get back to building. Visit deal.com slash sorcery. That's D-E-E-L dot com slash sorcery. But we do have another Alfred question. Yes. What is the craziest hack that you and Deepika pulled off? [45:27] story in my speech at Deepika's wedding. But the way that it goes is that we [45:32] We were partnered with a health system, and I won't say their name, and they had given us access to residual blood samples. Now, Deepika was like... [45:42] She's like a saleswoman. She's really good at like getting into situations, talking her way into things, out of things. [45:49] you know, [45:50] Our residual blood sample partnership with this health system [45:54] was actually nothing more than Deepika being friends with the security guard and the front desk lady. Oh my gosh. And so we had accumulated hundreds of samples and they were, you know, anonymized and, you know, there was no like HIPAA concern. But we were using these samples. We were getting them from the security guy? We were getting them because we would walk past the security guy. I guess he liked Deepika. And one day we, you know, I found out that, you know, this was the situation and

[46:21] we quite literally had to meet with, we were summoned to meet with the Health System Administration. And they were like, what the fuck is going on? Like, we were like 19 year olds. We were scared out of our minds. - You're like, what do you, this is bad? What do you mean? We're building a company here, don't you understand? - We're trying to make it work. And they were like, look, [46:41] Pharma buys these samples for like tens of thousands of dollars of samples. You guys are just picking them up. Like, what are we talking about? And, um, [46:48] Ultimately, we were able to use the fact that it had been going on for so long as a negotiating chip. And the worry that's their fault, like, you know, we just really turned it off and ended up having a very fruitful partnership. And it was an amazing it's an amazing academic health center, still a very close partner of ours. Just had a slightly unconventional beginning to that relationship. At the end, they're just like respect. Respect. No, it was it was very much like we get it. You're hustling and we support. [47:17] Another good Alfred question. Did you ever think you'd become an enterprise salesperson? You know, Alfred truly reminds me at every chance that he gets that I am now a sales rep. And that's okay, because I think sales is the most important skill in the world. And I think the best, you know, the best CEOs are always selling, they're always closing. And a big part of, you know, to Alfred's point, a big part of my job today is I'm an enterprise sales rep.

[47:47] I was at the Stanford AI lab. I thought I would be a hacker my whole life. I thought I would get a PhD. I thought I would go work at a research lab. And that would have been a fun career too, [47:56] This is just a very different set of skills. And I had to learn how to sell. It was not something that initially came naturally to me. But I'm so grateful that I took the punches and eventually got reasonably good at it. [48:09] Now that I'm saying these questions out loud, I'm really noticing a pattern. So here's the next one. He would like to know, did you think it was a good idea to show up to the JPMorgan Health Summit in a t-shirt, shorts and flip flops? [48:29] This is a, this story is turned into lore within the company because the first JP Morgan conference that we went to, [48:37] I did in fact show up in t-shirt shorts and like cargo shorts, like with the buttons and everything and flip flops. Wow. They were sandals. They were open toed, but they were Velcro. Were you 19 at this time? I was 20. I should have known better. Okay, all right. Yeah, 20. But the, you know, we go to the conference and I guess the cool thing at the time was you're a tech bro, you underdress, you go to these things. [48:59] And honestly, I'm at the JPM conference and these people are looking at me like I'm a fucking joke. And that was when I got my first suit. And so ever since then, anytime we go to JPM, I look good. And Alfred always comments on it as well. So I remember he saw me that day and he was like, this is ridiculous. This is completely unreasonable.

[49:20] And then he saw me like a year or two later when I showed up suited up. And that's when he started calling me an enterprise sales rep. So there's no winning with Alfred Lin. [49:29] So you can confirm you own a suit. [49:31] I own multiple suits now. All the colors, the best colors. That's great. So for Alfred's interview, [49:40] I think-- so Tariq introduced us probably over the summer. And then I was interviewing Alfred, and I asked you if you have any good questions for him. Yes. And you described Alfred as being-- [49:53] more of like a CFO, COO brain versus a VC brain. Yeah. So what is that like? What is it like working with him and Sequoia? [50:02] I think Alfred is so analytical and I think in some ways [50:07] the world has been shortchanged that Alfred Lin doesn't still run a company. And I mean, he gets to exert his leverage and advice across dozens or hundreds of companies now. But in some ways, he's such an incredible operator and he has such a great instinct for [50:24] how the roles of a COO and CFO overlap, because he did both of them. And I think what that means is when you show him a P&L, he will scrutinize [50:32] everything down to like the last item. Like why was TNE on sales and enterprise and the western segment so high this quarter? And you're like, [50:39] "Dude, what? What are you talking about?" And then you look in and you're like, "Well, actually, [50:44] they threw a couple big dinners for clients and we better see some ROI on that in a couple months. And so his instincts are very sharp. I think he is

[50:52] uh honed in on the numbers that matter in a business and he can read very very early the signals of [50:58] when you see traction and when you see great customer retention and when you see customer love in these products, I think makes him such a great investor. And then on the operating side, I think he's an incredible hirer and aggregator of talent. One of the biggest value ads we've gotten from Sequoia and Alfred is whenever we've had a candidate that we are right there at the finish line and we're competing with [51:20] a larger company or a larger comp offer, [51:22] Alfred comes in and gets the job done. And so he's a closer. He pitches the vision. He pitches the long-term bull case for the company and for me and Deepika as founders. And I'll say early on, [51:34] the massive difference that made, I mean, it was trajectory changing. There were executives, like, you know, amazing engineers at companies [51:43] who probably had no business joining us. Without Alfred putting in that word and vouching for us, they would have probably gone somewhere else. And now they're still at the company to this day. I mean, our CTO, Alfred was the closing call. Drew is still with the company and really one of the founding partners of the business. So I'm always grateful to Alfred for that. And to this day, he does it for us. [52:05] That's great. [52:06] Any other lore you have on your investors? What can we, I mean you're warmed up now. [52:11] lore. I mean, I got to position the missiles back. That's like a [52:17] Yeah, I would say about Heymont, I think one of the first time I met HT was I was in

[52:23] I was in college and I showed up [52:26] Because they-- didn't they-- wasn't this incubated with them? It was very early. So exactly. Camille was incubated at GC. And Othellis, which is what I was running when I first met HT, [52:38] I was starting it right out of my Stanford dorm room at the time. And so I went to the GC office, which is on University Avenue, and I rolled up and I had a bike and I rolled up and the old General Catalyst office in Palo Alto was like a house and then there was this white picket fence. And he was on the porch on a phone call. [52:59] And I was like, where the fuck do I put my bike? There's no bike rack. This is like, we're in Palo Alto, where are the bike racks? And so he's on the call and I'm just like, I had no idea who he was. I was like, hey man, is it cool if I lock my bike through this fence? And he was like, what? Who the fuck are you? And he was like, sure. And then he saw me walk in and we had a meeting later. But I think it just goes to show that the best investors are so plugged in that they're meeting... [53:23] with the random autistic kid on campus and giving them time a day. When, you know, HT was already running Livongo and he was already a big-time investor in Stripe. And the fact that he was giving me even like 20 minutes as a Stanford freshman just shows you, I think that's what it takes. I think to be great at anything, you have to give it so much time. And both Alfred, Hemant, Teresa, Kassar from Applied, who's also on our board. I mean, these guys are just always fucking on. [53:51] Before I ask you the last question,

[53:53] we're going to ask you a pre-last question. [53:56] Deal. What is your hottest take right now? I think the US health care system gets shit on a lot in that people say that [54:03] we have a bad health system and it's expensive and this and that. The reality is, is that the American health care system is the engine of innovation for the whole whole world, like the vaccines that are made are the best in the world. The drugs that are made are the best in the world. And I think from a [54:20] We owe it to the American healthcare system to make it more efficient. But this idea that you can get the same quality of care in Canada or in the UK is preposterous. I mean, you're waiting in lines for a basic primary care visit for a year in the UK. You're waiting. You're not going to get care. You're not going to get good long-term care in Canada, just period, because the systems and the infrastructure and the funding doesn't exist. And you have these 18 fucking provinces all never talking to each other. And so the American healthcare system is quite advanced. [54:50] with the operators that run it. [54:52] They're all really well-meaning. And I think they're... [54:55] has been 50 years of middleware that's gotten created with insurance companies and PBMs and, you know, all their interactions that there's a lot of profit seeking and a lot of rent seeking. But the quality of the system is probably the highest in the world. And now it's just a matter of [55:11] LLMs are this gift that we've received to go nuke all of that work tax. And that in some ways is going to be my life's work in terms of just eliminating the work tax and making the system [55:21] a trillion dollars more efficient. Because if you do that, you make the entire American middle class wealthier.

[55:26] As we look forward the next 12 months, what are you [55:30] most excited for. [55:32] I think the capabilities of these truly agentic models, we see them in R&D today. We see them even with our own engineers. And to see that now transition to the nurses and the physicians, it's like 2023 all over again, where the first time a physician adopted ambient scribing, their jaw dropped. [55:50] The first time a nurse was able to recite a prescription, and it all showed up perfectly in transcript as a structured note, [55:57] her jaw dropped. And this year, you're going to see agents where you give it a task in an EMR, and it just goes and does all the work, and dollars show up at the other end of it. And so, for me, I think the... [56:09] As an engineer, the thing you love is giving people these jaw-dropping moments, and we're at the precipice of another one in healthcare. So that's what I'm pumped for. [56:18] Wow, that's a great way to end it. Well, tonight, thank you so much. Thank you for sharing so much on Commir, the company, the history, everything that you've learned in the healthcare system. I haven't done many healthcare interviews yet, and it's really cool to see now the inflection points being had in that industry with AI. So thank you so much for taking the time, and congratulations. Thank you so much. Appreciate the time. It was a fun conversation. Hey, it's Molly. [56:48] a once a week top deals and tech headlines email and also go deeper on our podcast interviews. Subscribe to Sorcery today and don't forget to subscribe to the podcast on YouTube, Spotify, Apple, or wherever you listen. Link in description to sign up.