# Founder10x - Are AI Agents The Future of Doctor's Office? - Ganesh Padmanabhan, Founder, Autonomize AI

An episode of [VC10X - Investing, Venture Capital, Asset Management, Family Office, Wealth Management](https://invisiblenetwork.ai/podcast/1632806986), 21 Oct 2025, 58 min.

Audio: https://anchor.fm/s/a453ddd4/podcast/play/110010279/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net%2Fstaging%2F2025-9-21%2F409702444-44100-2-0726dddb5318e.mp3

Ganesh Padmanabhan is the Founder & CEO of Autonomize AI. After eleven years scaling infrastructure at Dell, Ganesh took on healthcare's biggest problem—an industry drowning in data but struggling to use it. We're discussing AI agents, healthcare transformation, and why patients believe healthcare has deteriorated despite all the technological innovation. ⭐ Sponsored by Podcast10x - Podcasting agency for VCs - https://podcast10x.com Autonomize AI - https://autonomize.ai/ Ganesh Padmanabhan on Linkedin - https://www.linkedin.com/in/padmanabhan/ Stories in AI Podcast - https://www.storiesinai.com/ 🎙️ Hosted by Prashant Choubey Follow Prashant on X - https://x.com/ChoubeySahab Follow Prashant on LinkedIn - https://www.linkedin.com/in/choubeysahab Timestamps: (00:00) Introduction (02:22) Ganesh's journey from Dell to founding Autonomize AI (06:55) Overview of what Autonomize AI does (10:03) Initial go-to-market approach and challenges in healthcare (14:51) Developing channel partnerships (18:15) Organic demand for AI in healthcare (23:07) Challenges of implementing AI in healthcare (data privacy, hallucinations) (29:17) Series A fundraising approach and strategies (36:16) Details of the fundraising process (40:24) Deployment of capital raised (42:45) Balancing rapid product development with lengthy healthcare sales cycles (46:27) Metrics used to show ROI to customers (49:59) Most pivotal decisions that changed Autonomize's trajectory (54:00) Playbook for building credibility in heavily regulated industries like healthcare (57:11) Closing remarks and ways to follow Ganesh and Autonomize AI For sponsorship or guest appearance requests, write to prashantchoubey3@gmail.com Subscribe to VC10X on YouTube, Spotify, and Apple Podcasts.

## Transcript

[0:00] 1/3 of the world's data is produced by healthcare and
[0:03] health data, right? It's more now, right?
[0:05] Yet we're as an industry, you're the furthest behind putting all
[0:09] that data to work. Last 10 years in the United
[0:11] States, one of the largest healthcare industry that the GDP
[0:14] in the world. You know, health tech as a
[0:17] investment category. I mean, you know, this health
[0:19] tech did well, health, IT did well, healthcare services did
[0:22] well. But if you ask the average
[0:24] patient has healthcare gotten better or worse, most of them
[0:28] will tell you it's gotten worse. If you think about it, the
[0:31] United States of 330 million people and we have less than
[0:35] 3,000,000, you know, physicians to take care of them.
[0:38] Healthcare cannot operate like the way it is done so far.
[0:43] Hey everyone, this is Prashanth and I'll be host for the VC
[0:46] Connects Podcast. And today we have Ganesh
[0:48] Padmanabhan with us. Ganesh is the founder and CEO of
[0:52] Autonomize AI. After 11 years of scaling
[0:56] infrastructure at Dell, Ganesh took on healthcare's biggest
[0:59] problem, an industry Downing in data but struggling to use it.
[1:03] We're talking AI agents, healthcare transformation, and
[1:07] why patients say healthcare has gotten worse despite all the
[1:10] tech innovation. So without wasting any time,
[1:13] let's dive straight in. This episode is sponsored by my
[1:16] own podcasting agency called Podcastenix, where I help
[1:20] venture capitalist, asset managers, allocators and family
[1:23] offices and starting their own podcast and building a strong
[1:26] brand. Podcast NX is designed in a way
[1:29] to work well with the busy schedule of investors where we
[1:32] handle every aspect of podcast production so you can focus on
[1:36] investing. If you're looking to build a
[1:38] strong investor brand, starting a podcast should be right up
[1:42] your priority list. It's the only form of marketing
[1:45] where you can showcase your strong network.
[1:47] In other words, access to limited partners and founders
[1:51] and grow that network further. And contrary to what most people
[1:54] hear, it's still very early to start a podcast.
[1:57] There are at least 10X more newsletters and thousand X more
[2:01] blogs than there are podcasts in this world.
[2:04] You can learn more about our podcasting services at Podcast
[2:07] ten-x.com. I'll make sure to add the link
[2:10] in the description below. Enjoy this episode.
[2:13] Hey Ganesh, so good to have you on the BC 10X podcast.
[2:16] How are you doing? I'm doing great, Vashan, great
[2:19] to be here and good to see you. Yeah, it's a pleasure having you
[2:23] on. To start things off, can you
[2:25] walk us through your journey from spending eleven years at
[2:27] Dell to founding Autumn Ice AI? What made you pivot to
[2:31] healthcare AI? No, thank you.
[2:34] Thanks for having me again, Prashanth.
[2:36] So my background is like, you know, it's, it's very, that's
[2:40] very vanilla, right? I grew up in, we were talking
[2:43] about it. I, I grew up in Kerala, did my
[2:45] engineering after I, you know, finished my 12th standard in
[2:49] mechanical engineering. Never really worked a single day
[2:52] in mechanical engineering, but got into mechanical engineering.
[2:54] But really understood, you know, I loved it because I loved
[2:57] machines and stuff at that time. Then I went into a, you know,
[3:00] the, the traditional corporate thing.
[3:01] I got a software engineering job.
[3:03] I used to work for a company called Adaptec, then Intel and
[3:06] then Dell. No weird sort of circumstances.
[3:09] Dell moved me from Bangalore. I was working in Bangalore to
[3:12] Austin, TX, which is where I live right now.
[3:14] And I did a lot of stuff at Dell.
[3:16] You know, it was like, I like to say Dell is where I grew up.
[3:19] So I grew up trying how to be an operator, how to actually scale
[3:22] a business, how do you start new businesses and stuff like that.
[3:25] And I was responsible for a Converge infrastructure
[3:27] solutions business, which we grew from nothing to $1.3
[3:30] billion in in revenue in five years.
[3:33] So at the end of that, you know, it was like right after the Dell
[3:35] EMC merger and stuff like that, I'd spend about 11 years to Dell
[3:38] at that time. And I was like, what's next for
[3:41] me? And you know, 2, two things
[3:42] happened. 1 is as any, as an immigrant, first thing that
[3:47] happened for me was like, I got my green card or permanent
[3:49] residence, which means I'm unshackled from working for
[3:52] somebody. I can do something on my own.
[3:54] Second thing was like, it was also the coming of age of, you
[3:58] know, big data and analytics and stuff like that.
[4:00] In 2015, sixteen AI was initially, you know, you know,
[4:04] surfacing up from the from the bottom.
[4:07] And then so I wondered, I started a company initially,
[4:10] which led to joining a, you know, a later stage company who
[4:13] acquired me to it, which was focused on explainable AI.
[4:17] So the the thesis of that time in 20/15/2016 was if most
[4:22] businesses are going to use AI, black box businesses don't, will
[4:26] not buy black box AI. Now little did I know what is
[4:29] happening today, but you know, started a company which was like
[4:33] a Jiffy Lube checklist for machine learning models.
[4:35] You bring your model, bring your data to that.
[4:38] So I did that, did spend 2-3 years there in that company at
[4:40] cognitive scale, Rande growth function, best job I ever had.
[4:45] And I learned from how do you go from big company operator
[4:48] mindset to builder mindset and go scale from, you know, zero to
[4:51] 1 and one to 10 kind of stuff. And then I did another start up
[4:54] after that, which is focused on data because, you know, in
[4:57] 2018-2019 we noticed this. Most AI projects will go to die
[5:02] in the data graveyard because all the data is siloed.
[5:05] It's all in different places and enterprises.
[5:07] So at Molecular, we were building a platform to
[5:09] centralized that data for enterprises to generate value.
[5:13] And then COVID hit and the whole world changed, right?
[5:16] So I left after our Series A race on that company and I was
[5:20] like, what do I do next? And it was a, it was very
[5:23] evident that one of the biggest problems that humanities,
[5:27] humanity faced was, you know, healthcare, like health, you
[5:30] know, we, we learned like you can have all the wealth you
[5:32] want, but you know, people in India, people in the US, people
[5:35] in, you know, different parts of the world were dying of COVID
[5:38] because, I mean, everything else falls flat if you don't have the
[5:42] health thing figured out. So I want to do something here.
[5:44] So spend a year consulting, learning a lot about the
[5:47] industry. I'd spend a lot of time in the
[5:50] healthcare industry when I in my previous companies, we really
[5:53] wanted to solve something that can a, create a generational
[5:56] company that we do things differently.
[5:58] So we started and autonomized with a very simple mission,
[6:01] right? Healthcare operations can be,
[6:04] you know, done a lot better. I mean, I like to say a few
[6:07] things like one, one is, you know, 1/3 of the world's data is
[6:11] produced by healthcare and health data, right?
[6:14] It's more now, right? Yet we're as an industry, you're
[6:17] the furthest behind putting all that data to work is if you ask
[6:20] anybody, right, last 10 years in the United States, one of the
[6:23] largest healthcare industry that the GDP in the world, you know,
[6:27] health tech as a investment category.
[6:29] When you know this health tech did well, health, IT did well,
[6:32] healthcare services did well. But if you ask the average
[6:36] patient has healthcare gotten better or worse, most of them
[6:40] will tell you it's gotten worse. That's what we're trying to
[6:43] solve with Autonomize AI. How do you create an
[6:46] infrastructure for healthcare to run better, better for patients,
[6:50] better for providers, and then just meaningfully just improve
[6:53] the quality of life for everybody on the planet?
[6:56] Absolutely. That's quite a mission and every
[6:59] no sort of fact. Healthcare is a massive
[7:01] industry. Of course everyone is impacted
[7:03] by health if like you said, that's the core of everything.
[7:06] If you don't have that figured out, everything else falls flat,
[7:10] right? So give us an insight into what
[7:13] autonomize AI does. What does it look like in in
[7:17] practice? Yeah.
[7:19] So look, I mean, autonomize AI, we, we started, you know, with
[7:22] this broad big mission, right, which is saying, look, you know,
[7:26] if you look at healthcare, there's a lot of inefficiencies.
[7:29] The cost of delivering care is very high.
[7:31] The inefficiencies are very as as Jeff Bezos says, your margin
[7:34] is my opportunity. So we looked at all the
[7:37] inefficiencies. If we can cut that
[7:38] inefficiencies, that's the opportunity to go do So.
[7:41] What we do at Autonomies is we built, we, we, we have built,
[7:45] you know, we have a library of AI agents that are focused on
[7:48] our capabilities that can help a healthcare enterprise run their
[7:53] operations more efficiently. I'll give you 2 or 3 examples.
[7:56] Prior authorization is a process that, you know, it's very, you
[7:59] know, it's, it's very prevalent in the US when it's a multi
[8:02] payer system. And so patients have to get
[8:05] approval or that their providers have to get approval for
[8:07] expensive drugs and processes from the health insurance
[8:10] company. To do that, they send a request,
[8:13] usually over faxes or electronic submissions to the health
[8:16] insurance company. They employ hundreds and
[8:19] thousands of nurses and doctors to review this, map it against
[8:23] clinical policy, make adjudication decisions and
[8:25] stuff. The problem with that approach
[8:27] is this one, it takes two or three weeks to hear back what
[8:30] their decision is on a request for a drug or something, which
[8:33] is, you know, crazy for a patient #2 it's extremely
[8:36] onerous for the provider who hardly is like there's, if you
[8:39] think about it, the United States, the 330 million people,
[8:43] and we have less than 3,000,000, you know, physicians to take
[8:46] care of them, right? And so it's a very, very, very,
[8:49] you know, you know, tight workforce and instead of working
[8:53] at the point of care, you have all of those folks employed at
[8:55] health insurance doing paperwork to review and map against
[8:59] different things. So there's another problem that
[9:00] we saw. So what we do is autonomize,
[9:02] like we, we built an AI system or an AI copilot for that health
[9:06] plan to go and automate the intake of that information, run
[9:11] it against multiple things. So it's truly agentic where it
[9:13] actually looks at, you know, where do I find this information
[9:16] goes, fetches that information from a system, adjudicates that
[9:19] provides a summary, all with the human in the loop.
[9:21] You give it to that particular nurse who's doing the
[9:24] validation. So they can do it in not 40
[9:26] minutes, but in 4 minutes or less, right?
[9:28] So that's one example. But similarly, there are
[9:30] hundreds of processes like this that exist across the enterprise
[9:34] and what we build as a library or a marketplace of those kind
[9:37] of agents that can Co create and code as you know, Co Co, you
[9:41] know, Co work with the human employees and the healthcare
[9:44] knowledge workers in a healthcare enterprise.
[9:46] So that's what we do, autonomize.
[9:49] Absolutely. Thanks for sharing that quite
[9:51] interesting share with us. What was the initial go to
[9:54] market approach that you tried and was there any sort of
[9:57] resistance to it because Healthcare is sort of has its
[10:01] own way of doing things and it can have resistance to change
[10:04] and so on. So give us an insight.
[10:06] Yeah, Yeah, I know it's a, it's a great question.
[10:08] And Healthcare is a go to market generally is hard and I think we
[10:11] can all align to that, right. I mean, like you can build the
[10:13] best product, you can build the best, you know, teams, but you
[10:18] have to figure out, you know, the business exists because
[10:20] somebody's buying, right? So we, you know, I'm a huge
[10:25] believer in a lot of experimentation.
[10:26] You experiment, you see what works and there is no silver
[10:29] bullet in this. Like what works for me and for
[10:31] Autonomize may not be the same that works for my competitor,
[10:34] right? What worked for us, how we
[10:36] actually approach this, our first customer we actually went
[10:39] into, you know, we knew that we will have a broad vision, but we
[10:42] want to go and you know, like prove it out and create and gain
[10:46] conviction in one narrow part of the segment that is large
[10:50] enough. If that is the only thing that
[10:51] we do, it should be big enough to do it.
[10:52] So we focused on biopharma and pharmaceuticals and, and you
[10:57] know, clinical research for them.
[10:59] So in clinical research, there's a process that happens wherein,
[11:02] you know, an average drug, when a pharma launches it, it takes
[11:05] about 10 to 15 years and two to $3 billion to launch, right?
[11:10] And 80% of the cost is in this process of phase two, phase
[11:14] three clinical trials where you're recruiting human
[11:17] recruits. You have to look at their
[11:18] medical history or to understand whether they're a good fit for
[11:20] it, whether they're going to have the right reaction.
[11:22] You use all the theoretical modeling that you did outside
[11:25] match it to actual physical, you know, patient data and then you
[11:28] recruit them and then you manage them to the trial.
[11:30] It's very painstaking. It's very risky sometimes for
[11:33] the patient. There's a lot of things that,
[11:35] you know, there's a lot of high, you know, hand holding that you
[11:38] need to go to make that happen. But it's very inefficient.
[11:41] It takes 10 to 15 years and you know, you're looking at
[11:43] recruiting 300 people. So what we wanted to do was
[11:46] like, we wanted to solve that problem.
[11:47] So my initial approach was like I reached out to everybody, our
[11:51] companies who are focused on the clinical research process and we
[11:55] very clearly said, look, we're you know, I have a team of
[11:57] engineers who can solve this problem, but we can we Co create
[12:01] this with you. So our initial customers were
[12:04] actually folks who when we go in and say we showed them the art
[12:07] of the possible, but we said, we don't have a product ready to
[12:10] go. We have to work with you.
[12:11] We are to Co create, Co build and Co create it.
[12:13] And that formed an ethos of how we went to go to market, right.
[12:17] And I'll tell you what happened right when, when that that, you
[12:19] know, turned out to be a great product.
[12:20] Most we had multiple customers taking that initial product that
[12:23] we had. We learned very quickly what
[12:25] they're trying to solve is not that patient recruitment
[12:28] problem. What they were trying to solve
[12:30] was inefficiency in that process problem, right?
[12:33] So then we pivoted or like we looked at it and say, instead of
[12:36] actually just being like, I give you a solution for every problem
[12:40] you have, I want to teach them how to fish.
[12:42] So we built a platform so they can build the different
[12:45] solutions on their own or we can build for them and stuff like
[12:47] that. And for that approach, again, we
[12:49] have to then find the first customer who was a large health
[12:52] plan who go in and say, look, we build these products you can
[12:55] use, but we also have a platform that it's in progress, be a
[12:59] design partner with us, right? We will we will give you
[13:01] preferential early access and stuff.
[13:03] So that kind of a product LED iterative at this what really
[13:07] worked for us from a go to market standpoint.
[13:09] The other thing in healthcare, like you said, Healthcare is a
[13:11] tough market, right? There is a lot of there's a high
[13:13] bar for quality. There is a high bar for, you
[13:16] know, people don't really all understand the, you know, AI
[13:19] very well. So you have to educate them.
[13:21] So you need trusted sources that you can actually get them to.
[13:24] So we did like early on in our life cycle, we unlocked channel
[13:28] partners who have the trusted relationships.
[13:31] We recruited a team of advisors who were, you know, who were
[13:34] respected. They were champions in their,
[13:37] you know, a time and they were actually running these
[13:38] healthcare enterprises and they really knew and understood the
[13:41] problem. And we, we know, we convinced
[13:43] them that look, this is the way you solve it.
[13:45] They became, you know, users and champions of it and then they
[13:48] became advocates. So we created a community of
[13:51] networks, a network of in our community of folks who are
[13:54] respected. And that's so it's not a single
[13:57] thing. But what we did in the go to
[13:58] market was like, and then we the last thing we would say is the
[14:00] only go to market side is like we, we don't go in and say, I'm
[14:04] going to sell you a platform that's going to take three years
[14:06] to implement and get everything done right.
[14:09] We will go to where the hardest problem you have today and most
[14:12] health plans, it's prior authorization, most, you know,
[14:16] health providers, it's around clinical quality or you know,
[14:18] intake or something like that. And then we go in and say, look,
[14:21] we'll solve that problem for you.
[14:22] Don't worry about the platform. The fact that we can solve it in
[14:25] three months and not 18 months like all our competitors will do
[14:29] is because we have a platform. So they see the value
[14:32] immediately. We give them value very quick
[14:34] ROI in the next first two to three months and then right
[14:37] after that say, how are you able to do it so quickly?
[14:39] Well, we have the broader platform in the library of
[14:41] prebuilt assets that we used to go do it.
[14:43] Oh, then they open up the door to say let's solve problem
[14:46] 23456. So it's a land and expand sales
[14:50] motion for us. Absolutely.
[14:52] Quite interesting. You mentioned how design
[14:55] partners was initially approached it and then you
[14:57] unlocked channel partners. So at what point in the journey
[15:02] of the startup where you say did you say that, OK, now we're
[15:05] ready for channel partners, let's start unlocking that that.
[15:09] Yeah, I think, you know, let me, let me rephrase it.
[15:13] It's like it's less full blown channel partners in this size
[15:16] more about resellers, right. So what we did was early on, you
[15:21] know, we, we didn't want to be like, you know, so there's 22
[15:26] phases of, you know, your partnership build out, right.
[15:28] So one phase is the initial phase is like, you know, you
[15:31] have a product, you know, you can deliver very quickly, but
[15:33] you don't have access to the customers.
[15:35] So how do you solve for access? That's what we did first phase
[15:38] one. And that what we did for that is
[15:40] like, we actually looked and mapped out the partners who are
[15:43] not your biggest and the most popular on the accentures and
[15:47] the folks of the world. Now we have partnerships with
[15:50] them too right now. But at the time we look for
[15:52] folks who were like didn't have the skill set.
[15:55] They didn't know that they can actually do it.
[15:56] But they, the customers are asking, they were very good
[15:59] customer relationships and that customers were asking them
[16:02] saying, Hey, can you solve this with AI?
[16:03] And they're like, we don't know AI, maybe we can refer somebody
[16:06] else to you, right? So we identified them and
[16:09] incentivize them well enough. Like they, we shared a lot of
[16:11] revenue with them to go in the door and say, like, we'll prove
[16:14] it to you. And they were all worried about
[16:16] like, hey, quality should be good because we have a
[16:18] reputation at stake. So we knew we had to optimize
[16:20] for that. And so we delivered high quality
[16:22] output and we gave them value. So they became fast and then
[16:25] they started introducing us to other customers and stuff like
[16:27] that. So that's one part of it.
[16:29] The second phase of the channel partnership is what we're embark
[16:32] on right now. It's like how do you then go and
[16:35] really build out a distribution network more than just like
[16:38] friendly partners will refer you and that involves a multi tiered
[16:42] approach, right. One is technology partners.
[16:44] So we are now available like the the Co pilots from autonomous AI
[16:48] is on the Microsoft Azure marketplace.
[16:50] So we have a coastal partnership with Microsoft, they go in and
[16:53] introduce us to customers. Now they won't do the selling
[16:55] for you, but the customers know their sellers get paid when we
[16:58] sell our products, all of that stuff.
[17:00] The third tier of that is then the another, another part of
[17:03] that, that's the technology partnerships and there's the SI
[17:06] partnerships where we're actually enabling organizations
[17:09] like Deloitte and Accenture to build whole business unit on top
[17:13] of our platform, right? And you know, eventually.
[17:15] But today what they're doing is, look, I have a problem that the
[17:18] customer's asking for if I build my own technology team to go
[17:21] build them by the hour and do it, it's going to take me 18
[17:24] months to get there. You guys have a ready to go.
[17:26] We can do the implementation services.
[17:29] Let's go partner. And that's with your, you know,
[17:31] emphasis your cognizance of the world.
[17:33] So all these size, we, we are very partner friendly.
[17:36] We know that we build an infrastructure and enablement
[17:39] platform and they make it come to life.
[17:41] So there's a win win in this process.
[17:43] And so that's the next year that we're actually embarked on.
[17:45] We're enabling this partnership ecosystem that requires a lot
[17:48] more investment, right? That requires you to build out a
[17:50] training function, you have to build enablement modules and
[17:55] it'll make your product a lot more self-service and all of
[17:57] that stuff, right? We also do today a lot of
[18:00] forward deployment engineering. So you know, when we see the
[18:03] gaps that our partners cannot fill, we bring in our engineers.
[18:06] That's not a primary business model, but we do know that
[18:08] that's actually something relevant for getting the
[18:11] customers really you know that white glove service to go to get
[18:14] value with AI. Absolutely.
[18:16] And from what you're saying, it sounds like there is an organic
[18:20] demand for AI in healthcare versus from what I assumed that
[18:24] it would be a sell at this stage to, you know, bring AI into
[18:28] healthcare organizations. What's your take on that?
[18:32] Yeah, no, you're, you're right. I think.
[18:33] I think it's a push and a pull, right.
[18:35] So the pull, it happens because everybody is like, look,
[18:38] healthcare cannot operate like the way it is done so far.
[18:41] I mean, humans cannot operate like the ones that like the way
[18:43] we have. The amount of information we
[18:45] take as a human being has far exceeded the the capacity and
[18:49] the capability of our brains. So you need an augmentation
[18:51] layer, you need the AI, you need the Jarvis suit to go make that
[18:55] happen now. That is 1000 times worse in an
[18:59] industry like healthcare where like, you know, six years ago
[19:02] or, you know, 1020 years ago, the only data you needed that
[19:06] you couldn't get access to, to make a decision on a patient's
[19:09] care plan was what they told you.
[19:12] Maybe there was no research, there was no this thing and
[19:14] stuff. And that's also why the, the,
[19:16] the, the, the quality of life, or rather the, the life
[19:19] expectancy was much lower 100 years ago versus not right now.
[19:23] You have all this information, but then it's the same human and
[19:26] the same brain and the same teams and the same structure.
[19:28] You cannot process it all. So the organic demand happens
[19:32] because I think healthcare executives and the organizations
[19:34] are realizing that the only way they can get away out of the
[19:38] cycle of margin compression, being able to deliver service,
[19:42] being able to extend the reach of care, right?
[19:44] So today access to care is a huge problem around the world.
[19:47] And that's because we're not producing enough MDBSS or MD's
[19:51] or, you know, folks who actually are in medicine and they're fast
[19:54] enough, right? The education takes long,
[19:56] there's a lot of practice. So you have to augment them.
[19:59] So I think there's a big real. Realization that AI is
[20:02] potentially A savior. Now, I'll also caution that
[20:05] there's also a lot of fear and apprehension on the healthcare
[20:10] industry because I mean, the cost of mistake is way high.
[20:14] You know, you, you know, your, your, your Facebook Messenger
[20:17] goes down for three days. Nobody will miss it.
[20:19] The people complain, you know, bitch and moan about it a little
[20:22] bit, but they won't miss it. Your billing system goes down
[20:25] like it did for the United hack that happened a year ago.
[20:29] It's being it brings livelihoods down, right?
[20:32] So it brings lives down. So I think there is a lot of,
[20:35] you know, concern of that. So even though there's a lot of
[20:37] demand, there's a lot of noise in the system.
[20:40] You know, you hear a lot of start-ups coming in and saying,
[20:43] OK, we'll give you an agent that will just solve everything.
[20:46] It'll just completely automate all the things you were doing
[20:48] for your nurses doing. It's a lot more nuanced than
[20:51] that. In healthcare, there's an agent
[20:53] requires access to all the information.
[20:55] Half the time the information is not in a digital form.
[20:58] In healthcare, when you're stumbled with a decision on how
[21:02] to watch payment policy, you adapt to it.
[21:04] You call Sally, who's sitting on the other room.
[21:07] And if Sally's not available, you know, you wait for that
[21:10] case, right? How do you capture that in poor
[21:12] agents? So there's a lot of nuance in
[21:14] that front. And So what we've done is
[21:16] actually engineered a system that, you know, allows you to
[21:18] hand off between AI and humans. So those are the nuances, right?
[21:22] Where in, you know, when you deliver information to a human
[21:24] being or a nurse or MD, you want to give them citations, results,
[21:29] all of that kind of stuff so that they can make their
[21:31] decision with less amount of time and with more context,
[21:34] right. And then on the other hand, when
[21:36] a human is like, you know, I have to look at this five
[21:39] different databases to find out an information and rat in and
[21:43] and reconcile this across this thing.
[21:46] And I've done it. I've done it 50 times and I have
[21:48] the age and watching how I do it.
[21:50] Just automate that whole thing for me.
[21:52] I don't want to do it anymore, right?
[21:53] So understanding which is the right context to automate, which
[21:57] is the right context to augment itself is a superpower in the
[22:00] way we think about it. So there's a lot of demand, but
[22:03] there's also a lot of noise. So right now that the market is
[22:06] about like from a go to market perspective, it's a combination
[22:09] of consultatively helping the map out what problems should you
[22:12] solve first? Everybody tells you, oh, let's
[22:14] just solve X and we go in and we map out this thing for them and
[22:17] saying, look, you're spending, you know, 60 FT ES doing this
[22:21] thing today. And you know, and then you have
[22:23] this other problem here that you're spending 200 FT ES and
[22:27] they're all frustrated. They don't want to work on this
[22:29] right now. That's a bigger problem that you
[22:31] should solve than this one, right?
[22:33] And customers appreciate that because you know, we're not, you
[22:36] know, one of the advantages that we bring in autonomize is
[22:39] because we have a broader platform.
[22:41] We don't go in and shove the customers that you shall use
[22:44] this, right. We give them there's a library
[22:46] of capabilities and there is a consultant.
[22:48] I mean, obviously we know where the big problems are.
[22:50] We look at a 10K statement from a, you know, from a health plan.
[22:54] We know your utilization is high, your quality heated scores
[22:57] are actually low. So we'll go and focus on that
[22:59] because that's going to be the immediate bang for your buck.
[23:01] So a lot of that consultative selling, but also the fact that
[23:04] there is a lot of demand and and pull from the market helps quite
[23:07] a bit. Absolutely, quite interesting.
[23:10] And I was coming to the concern with which you mentioned in in
[23:13] your answer. So what is that?
[23:15] Like you said, a lot of the data is in the physical form right
[23:19] now or in the minds of specific doctors on nurses, right?
[23:22] So the first layer of solving that becomes bringing that
[23:25] online in some way, right? And when you decide to do that,
[23:28] get approvals to do that, the next layer of concern becomes
[23:32] data privacy, because this is healthcare data, patient data,
[23:35] You have to handle it with care. You know, if you have, if it
[23:38] happened to solve that, then the problem become becomes that what
[23:42] if the AI that's doing all this hallucinates, right?
[23:46] The AI that we work with today tend to hallucinate sometimes,
[23:50] right? So that risk becomes much bigger
[23:52] in the case of healthcare, like you said.
[23:55] So how? How do you tackle all these
[23:56] three layers of problems? Yeah, No, I think, you know,
[24:01] there's, there's a few things, right.
[24:02] One is one of the most critical things in that whole spectrum is
[24:06] like, you know, even like it's hey, it's not really easy to
[24:09] just turn a data or a context or a knowledge online.
[24:13] I mean, because the fact is the reason, like there's every
[24:16] organization in healthcare, for example, they have their
[24:18] standard operating procedures, they have their job aids for new
[24:21] employees. They've all done a lot of work
[24:23] in there, but there's still gaps in that information because
[24:26] those are not easy to update and keep updated and stuff.
[24:28] So one of the approaches we do is we have an underlying
[24:31] knowledge craft that actually not, you know, it captures all
[24:35] the we declared things that the people say, here's what is
[24:38] actually happening. Here's what I want the inferred,
[24:42] you know, characteristics of a particular workflow.
[24:44] For example, it's like, look, every time you know, Sally gets
[24:48] these three inputs, it's never there in the job 8, but she
[24:51] always has this four thing, right?
[24:52] And we have ratified her across 5000 transactions is exactly
[24:55] what's happening. Well, that's an inferred
[24:57] characteristic. That's a probabilistic
[24:58] connection. And then there's observed,
[25:00] right? What does Sally say she does
[25:03] versus what she does? Right?
[25:05] There's a delta in that. So you're capturing all of that.
[25:08] So that becomes like a, you know, you're, you're creating a
[25:11] knowledge graph or a knowledge system for the enterprise that
[25:14] beyond autonomize. If they companies start to
[25:17] decide stop working with autonomize, they still are left
[25:19] with that particular knowledge. We would capture it for them and
[25:21] they can use it for their own. Now that's one side on top of
[25:24] this, I think what we do is like, you know, across our
[25:26] stack, we talked through this and we call it our pyramid of
[25:29] trust, right? And the pyramid of trust starts
[25:32] in three tiers, right? There is a tier of, you know, we
[25:36] call it guard gauge and govern, right?
[25:39] And the idea is like, so there's a guard layer, which is
[25:42] basically all your things, your basics.
[25:44] You want to make sure that you're maintaining provenance of
[25:46] where the data came from, what kind of data it is, who has
[25:49] access to it. The usual, the the usual
[25:52] encryption, your SHA trophy 6, your ACS, all of that stuff,
[25:56] right. So we're HIPAA compliant, all of
[25:58] that. So that's your first layer.
[25:59] The second tier is actually understanding context.
[26:02] And the context here is actually saying, look the a visiting
[26:07] physician looking at medical records.
[26:10] You know, one of the things that on HIPAA is actually, you know,
[26:14] authorized for use kind of framework, which is basically
[26:17] HIPAA prohibits you to use data for an unintended purpose.
[26:22] That actually how you can use it for right?
[26:24] And how do you capture it? There's no system to do that.
[26:27] It's only like what people tell you, right?
[26:28] But we, what we have is because we have a context engine, we
[26:31] understand the context in that the data is used that allows us
[26:34] to bifurcate and say like this is the context necessary for
[26:37] utilization management. This is the context necessary
[26:39] for, you know, care management versus claims versus that.
[26:42] So that kind of built, built brings in a distillation layer
[26:45] across that. The third thing is the
[26:47] governance, right? And the governance comes from
[26:49] two things. One is your classic thing,
[26:50] right, your hallucinations. So we have a lot of, you know,
[26:53] you know, LLMS Judge or other models that will actually go
[26:56] look at the models and make sure if you're hallucinating, you're
[26:59] going away from the core training data set.
[27:02] It'll flag that market for human review.
[27:05] There is a human in the loop for all critical clinical workflows
[27:09] that actually brings up, hey, I'm not able to make a decision
[27:12] or there's comprehend there's issues here and here's the
[27:15] evidence, here's the plus and minus and they get to make that
[27:17] decision and that what they do that gets encoded back in as a
[27:21] reinforcement learning into the system.
[27:23] And then there's a governance layer on like how is the model
[27:26] performing against the train data set?
[27:28] How is the model performing in terms of hallucination, the
[27:30] retrieval score, accuracy, F1 score, all of that kind of
[27:34] stuff, And also the business metrics?
[27:36] Because more often than not, what people don't realize is
[27:40] you're often not solving a technical problem in the
[27:42] workflow, right? You know, and I, I, I like to,
[27:46] one example we hear is like, hey, is your accuracy of
[27:49] extraction 99.9%? And there's a lot of problems
[27:54] with that question, right? If you think about if you
[27:57] actually measure it, the extraction accuracy from forms
[28:00] to typing up information for humans is at best 60% accurate
[28:04] and best, right? Then you know, there's, you
[28:07] know, and then there's also the integrator variability, which is
[28:10] like a well trained person versus a new person, all of that
[28:14] kind of stuff. Interpretation, eyesight, there
[28:16] are a lot of variables in there, but then there's also like when
[28:20] you think about 99%, you may not need 99% to get the value you're
[28:24] trying to do, right. You may not even need that 60%
[28:27] because what you're trying to the problem, the opportunity
[28:30] with AI is you don't have to solve things in absolute.
[28:33] All you need to do is actually, let's say you have, you know, 50
[28:36] headcount today allocated and doing like wasting their time
[28:40] doing data entry and validating against multiple databases.
[28:44] And even if you're able to take half of that volume and automate
[28:48] that, that is still a win. Now you could be doing that with
[28:52] 99% accuracy versus 50% throughput rate doesn't matter
[28:56] because you're getting, you're optimizing for the business
[28:58] governance lay right. So governance to me is not just
[29:01] the model governance, but it's also like, what does your agent
[29:03] do? How is it performing to goals
[29:05] and stuff like that? So we build an infrastructure
[29:07] around it. So we call this our you know
[29:09] pyramid of trust that we actually built in and we baked
[29:11] in across all our agents, all our infrastructure and and
[29:15] deliver that value for a health enterprises.
[29:17] Absolutely. Quite interesting.
[29:19] And you've recently raised about $28 million, CDC, so many
[29:24] congratulations on that. Do do tell us your fundraising
[29:29] approach and what were the strategies for targeting the
[29:32] right investors for you? Yeah.
[29:35] So we had AI mean we're very grateful for the for the the A
[29:38] round that we put together and we hadn't raised a lot of money
[29:41] before that. So we raised the 4 million C
[29:43] round back in 2023 and we still had a lot more money left in the
[29:48] bank when we raided the Series A race.
[29:50] I think First off in fundraising is has been around like two
[29:53] things. One is, you know, timing and
[29:55] targeting, right. I like to simplify this like in
[29:58] timing was perfect. The business was actually
[30:01] breaking out. We can see that we were added
[30:03] like in Q3 of last year, you know, outside of for early
[30:06] customers, we added one customer, Q4 we added two
[30:09] customers, Q3 we added three customers.
[30:11] So it's like the, the, the trajectory was solid.
[30:14] We were beginning to start seeing, we waited for the
[30:16] fundraise for Series A because we didn't need the money and be
[30:19] in healthcare results matter, not just the traction, right?
[30:24] So our early customers, it took us almost 6 to 8 months to start
[30:28] seeing good ROI for them. And they started becoming
[30:31] champions. They start talking about it and
[30:32] stuff. And that led to, they introduced
[30:34] us to some of our new customers. And so that momentum was
[30:37] building and the timing was perfect for us to go and say,
[30:40] look, we can show that we build a great business.
[30:43] It is right in the right trajectory.
[30:45] And these days ABC, all these numbers don't really mean
[30:47] anything because like, you know, we're building a scale around
[30:50] right now, right? And it's like Series A typically
[30:52] is a 1,000,000 AR. We were like way ahead of that
[30:55] when we did the Series A, right? And so 1 is like the timing was
[30:59] perfect. That tail windows in there.
[31:02] We also like, you know, again, I, I'll, you have a lot of VCs
[31:07] listen to your thing. I don't want to just go and
[31:09] raise money for the sake of money.
[31:11] So it's not for me, it's not about like personally, it's not
[31:14] about like how much money you raised and how soon you raised
[31:17] and stuff like that. I mean, those are probably
[31:18] relevant for the right reasons. But for me, it was very
[31:21] important that we build a sustainable business, that
[31:23] customers are happy, we're growing the business you're
[31:26] building good fundamentals, good economics in the business unit
[31:29] economics and stuff like that. And that we had a good shape of
[31:32] the business of the timing was right to raise money.
[31:35] Now second thing was targeting not everybody understands
[31:38] healthcare, right? We had, we are we had a very
[31:41] fortunate fundraise process start to finish.
[31:44] It took us two term sheets about four or five weeks, you know,
[31:47] maximum four or five weeks, which is very, very grateful.
[31:51] Again, part of that was actually because targeting was right.
[31:54] We talked to in the first meetings, I was telling VCs,
[31:56] this may not be a fit for you, right?
[31:59] Because if you don't understand a few things when we started the
[32:02] business and right now we have multiple customers and stuff.
[32:04] And it's like people who understand the industry, they
[32:07] get it. Healthcare is a concentrated
[32:09] market. Healthcare payers where we had
[32:11] the early traction was even a more concentrated market.
[32:14] So when the moment AVC says, but all your revenue, 60% is coming
[32:19] from this 2 customers, I'm like that's by design because that
[32:22] tells me that they're expanding, right?
[32:24] They're not just starting on the initial 3 to $500,000 ACV, but
[32:28] they're now done 5678 use cases and either you get it or you
[32:33] don't get it, right. So for us it was like targeting
[32:36] off, you know, getting the right help, you know, healthcare focus
[32:39] investors was important for us. We had a lot, we had like 5
[32:43] different term sheets and we selected what we were really,
[32:46] really happy and with who we selected as a lead.
[32:49] We try to fit in a few of the others into the round too.
[32:52] And something didn't work out and you know the size of the
[32:54] round and stuff too. But I think, you know, timing
[32:57] and targeting were the two big things on fundraising.
[32:59] And then I also believed in like making it efficient for the
[33:03] investors. So not just for you, right.
[33:04] And you know, we did a lot of things that you know, we did
[33:06] very well. We actually, I didn't bring in
[33:08] any of my, not a lot of my leadership, my Co founder was
[33:11] involved a little bit in the process, but I, I shielded them
[33:13] because we had to continue to build a business.
[33:15] So we made them, you know, record videos of overviews of
[33:18] their things and stuff. And once you're closer to the,
[33:21] to the, to the term sheet stage, I'll get them involved.
[33:24] And we ran, we, we basically led the process with them.
[33:27] I mean, there are some investors which still wondered what I
[33:30] mean, like we were, you know, our lead did an amazing job of
[33:33] going really deep into the market.
[33:35] They did a lot of market work, work that they shared with us
[33:38] after. So we've become, you know, even
[33:40] before we were like, look, even if you know, they're not the
[33:43] ones we do, I want to figure out a way to work with them because
[33:46] you get to know investors, the vouchers team by in the process.
[33:49] So we learned that. So we ran a process wherein,
[33:52] look, we may try to make it efficient and everybody
[33:54] respected it because, you know, most VCs will ask you were like,
[33:57] what's your process for doing this?
[33:58] And you know, you just had to be, look, it's very simple.
[34:00] I don't, I'm not going to put a deadline and a timeline and
[34:02] stuff, but I want to make it. I want you to get the conviction
[34:05] as soon as I can, right? Here's the results.
[34:08] What really helped us honestly was the business was strong.
[34:11] So when they look back at the business and like, oh, you're
[34:13] not just selling dreams, you, you're a great approach.
[34:16] You're very differentiated. You guys have built out the
[34:18] business. There's actual customers when we
[34:21] were raising and even today people who claim our
[34:25] competitors, we don't run into any of them in our, we are at
[34:28] three of the five top health plans in the United States in
[34:32] production for multiple, I mean multiple quarters now and some
[34:36] of them for multiple years, right.
[34:38] Nobody can claim that today. So it was like easier to sell
[34:41] from that perspective. So then it was about like you
[34:43] have a business, there's a strong thing, the product is
[34:45] strong, you've done is this the right thing team that I want to
[34:48] back? Can they scale?
[34:49] Can they this thing? So we understood the psyche of
[34:52] that the investors, the right investor, they were thinking
[34:54] about it the right way. And so we're very fortunate.
[34:56] Valtteris lend around Cigna Group ventures joined us as a as
[35:00] a follow on our existing investors came in with the asset
[35:04] management and ATX venture partners.
[35:05] Tal Ventures was an investor I always wanted to have on my cap
[35:08] table from the seed round. So I reached back out to them
[35:11] saying, hey, you know, mid, do you guys want to come back in?
[35:13] And they were like, yes, we're, we're in.
[35:15] So we'll do our regular seed check and like come on in.
[35:18] So we did that. We had some high net worth, you
[35:20] know, industry executives also we accommodated for the round,
[35:22] but it was a, you know, good sized round and helps us
[35:25] actually really deploy capital and scaling the product, scaling
[35:29] the forward deployment and customer success side of the
[35:31] house because we've learned that you have to help customers.
[35:35] It's not just like I can give a product and walk away and they
[35:37] can figure it out, right? The eyes new, the work flows are
[35:40] new. There's all these organizations
[35:43] are going through a transition moving from because most work
[35:46] flows in healthcare will build about 30 years ago, most work
[35:50] flows. How do you bring that to the age
[35:52] of AI? You have to reimagine it,
[35:54] rethink it. How is an AI first looks like?
[35:56] And that requires a lot of hand holding.
[35:58] So we're investing in all of that forward investing in all of
[36:00] that. So that was the process that we
[36:02] ran. And so we were very, very
[36:04] fortunate that, you know, the market was the timing was right
[36:07] and ran a pretty, you know, efficient process for everybody
[36:12] and very, very happy with the the final investors that came on
[36:15] the cap too. Yeah, that's that's pretty
[36:18] commendable. But a four to six week is very,
[36:20] very soon for for around to complete, right?
[36:25] Usually doesn't happen even if an investor has conviction, they
[36:28] would still like to wait and watch that.
[36:29] OK, I'm going to just sit and watch them for two to three
[36:32] months, see how it plays out to other than investors coming in.
[36:35] And then maybe I make a commitment, right.
[36:37] So there has to be some sort of, you know, leading them into a
[36:41] deadline or something for for you to close that soon.
[36:43] Yeah. I think it was, it was two
[36:45] things, right? One is, you know, yeah, so we
[36:48] set some deadlines, but it wasn't strict.
[36:49] I mean, I didn't tell any investors saying, oh, you have
[36:51] to make a decision by next week, right?
[36:53] Once I got the first term sheet, I did tell them guys I have to
[36:56] make a decision in a week. So if you're in or out, we're
[36:59] we're very happy with the first term sheet that we got.
[37:02] But I think we would love to have you as part of the Round 2.
[37:04] So make your best offer, right? So we did that, but the two.
[37:07] Things that was the lead the first term sheet, Yeah, the lead
[37:11] the. Yeah, the lead term term sheet,
[37:13] right. So the, and then we also didn't
[37:15] want to when we were racing like we were, we started off racing
[37:17] let's say 20 to 25,000,000. We didn't want to do someone
[37:21] leading us with a 5 to 6,000,000 round said, look, you're going
[37:23] to put in about 15 to 20 million to read the round because I
[37:26] don't want to go around chasing the rest of the money with the
[37:28] follow money, right. So what we did was a two things.
[37:32] One is I think more importantly building the process.
[37:34] We kicked off in mid to late February and we announced that
[37:38] we we closed in March, sorry, May first second week.
[37:43] We return sheets down in April and closed in May.
[37:46] It it towards the end of May I think.
[37:48] And then we announced in June, right, because we had the 3rd
[37:51] and 90 days to do it. So the big thing that happened
[37:53] was like when we started talking to investors in February and
[37:57] they saw the business in Feb and March.
[37:59] In April we had almost doubled the business during the process.
[38:04] So, and they could see that they were like, no, we closed the
[38:07] year. We, we doubled the business in
[38:09] Q1 for literally from last year to this almost doubled it.
[38:13] And they could see that trajectory closing.
[38:15] So I think one of the things was like, you know, you know,
[38:16] investors like to come in early, right?
[38:19] So we were like, look, I mean, we can race now or we can take
[38:22] the can down the road. And I was sure like, if I
[38:24] couldn't race, I couldn't get a lead in about four to six weeks.
[38:28] I was going to shut down the process and race again after we
[38:30] had more traction because I don't need the money right in
[38:34] that said, I think, you know, I knew that I could deploy capital
[38:37] if I have to be honest with you, right?
[38:39] So what really helped build conviction for a lot of
[38:42] investors is like especially our lead and they spent a lot of
[38:45] time with potential customers, our existing customers and
[38:48] stuff. And we also protected them.
[38:50] We made customers record video interviews so they can actually
[38:53] see it, but then not talk to the customer until they have closer
[38:55] to conviction. And all of that really held
[38:58] because they saw a lot of we got to know these investors during
[39:01] this. And you know, I mean, you, you
[39:02] know, Prashandi, you will know who you want on your cap table
[39:06] when you engage with them, right?
[39:07] I mean, so we kind of knew it was like, even though it's not
[39:10] the best offer, we want this investor to be on the cap table.
[39:13] And you know, we ended up taking on probably just the best
[39:16] valuation offer, right? We took the best overall offer
[39:20] and then during the process, how thorough they were, how deep
[39:24] they were actually going, how you could really see their
[39:26] understanding of the business will ensure that they can
[39:30] actually help us grow. The business was very important
[39:32] for us, right? We wanted people from the
[39:34] industry and stuff too. So I wouldn't say we didn't set
[39:37] deadlines per SE. But then once we got the initial
[39:40] term sheet, we told everybody, look, we get it's not fair on
[39:43] the the term sheet we got and say we have to go back to them
[39:46] at least. And it's only fair that we have
[39:47] to go back to them within a week, right?
[39:49] And then that time, you know, accelerated the timeline for
[39:51] everybody. But it was more like, you know,
[39:53] we saw the business in January, February, and then when they saw
[39:57] the business in March, it was like, Oh my God, what are you
[39:59] guys doing? So I want to get in early rather
[40:01] than wait for, you know, a much bigger round later this year,
[40:05] right? So that's that's the way.
[40:06] I think that happened, but yeah, you never know.
[40:11] Yeah, amazing. A great process there.
[40:14] And talking about deployment of the capital Grays, you said you
[40:18] didn't need the money at that point, but you still went ahead
[40:20] and raised that because you know you could deploy it in a good
[40:24] way, right. So give us an insight into how
[40:26] you plan to deploy that capital, What percentage of capital goes
[40:29] into what area of the business? Yeah, Yeah, sure enough.
[40:32] So I think look, I mean I would say like the way I characterize
[40:35] that statement as saying, but we didn't need, we were not running
[40:38] out of cash to go race around. So that gives us the cushion to
[40:41] go do it. We could actually get through
[40:42] the rest of the year with the cash we had because we are
[40:44] paying customers who are like we're very good at our financial
[40:47] discipline collections and all that stuff.
[40:50] But how we think about deploying this capital, so we there's a
[40:52] sizeable Series A round and we could have done more would be,
[40:55] you know, it's like what you'd races, what do you need The 3
[40:59] buckets of big deployment 1 is definitely double down on R&D
[41:02] and product. And that element of what we are
[41:05] actually deploying on that is, you know, I would say probably
[41:08] 40% is going to go towards product 40 to 50% I think say
[41:12] it's going to go to its product and the other 50% is go to
[41:15] market and customer success enablement, right, the
[41:17] infrastructure. So do that the big focus on the
[41:20] product is like we want to enable that next ecosystem of
[41:24] partners to build capabilities. Our customers are beginning to
[41:28] build on our platform which is amazing.
[41:30] So investing more in self-service, investing more in
[41:32] governance, investing more in trust and safety, things that
[41:36] you know, we know are very critical, like you can solve AI,
[41:39] you can build a model today, you can build an AI agent today.
[41:41] It's not that hard. But doing it in healthcare
[41:43] requires you to understand the regulations, the compliance, all
[41:46] of that stuff. So doubling down on that whole
[41:48] layer of governance and compliance is a big part.
[41:50] Even things like building an evaluation data and nobody has a
[41:53] way to evaluate whether an AI is actually working well or not.
[41:57] So providing that kind of harness and infrastructure for
[41:59] them is going to be huge. The go to market side of things
[42:02] going to be two big areas, right?
[42:03] One is classic sales and marketing build out like in
[42:06] terms of this thing. But the other big thing is like,
[42:08] you know, enabling customers, consultants who actually are
[42:12] clinically astute, who understand clinical, you know,
[42:15] economics, who are, you know, MD's or you know, the two
[42:19] medical officers building out the team that can help customers
[42:23] not just implement solutions, but drive the change management
[42:26] needed, make sure they can adopt it, understand and learn from it
[42:30] and basically Co create with them how the future needs to
[42:32] look like for the health enterprise.
[42:34] So that's the big, you know, 2 buckets.
[42:36] So I would say 5060% forty, I would say 50 to 60% on R&D and
[42:40] the rest 40% is more commercial go to market and customer
[42:44] success. Right, makes sense.
[42:46] Thanks for sharing that. How do you balance rapid product
[42:49] development with the Lendy sales cycles typical in healthcare
[42:53] enterprise deals? Yeah, it's a, you know, it's a
[42:56] good question, right? So you, you only build what the
[42:58] customers want, the number one, right, basic principle, right.
[43:02] There are things you're to build ahead.
[43:03] So, you know, for example, and I think there's a couple of
[43:06] companies who did this well, I mean, Dev Rev, they responded,
[43:09] did this really well, which is like, you know, when you you
[43:12] often don't know because in the world of AI, the problem space
[43:15] itself is emerging, right? What is the problem today is
[43:18] going to be not that big of a problem two days, right?
[43:20] So one thing we did was we innovative first on
[43:23] infrastructure, we build a platform.
[43:25] It's very counter intuitive, right?
[43:26] Which is like any use case you use will require you to do 1234
[43:31] and five. And we know that right?
[43:33] Build that you put it on a road map, you keep executing to it.
[43:37] What it's problems you're solving, the copilots, the
[43:40] agents, that stuff that is organically built, usually Co
[43:44] created with customers, right. We go in and say, look, we have
[43:47] a point of view and Healthcare is highly, you know, diverse 1
[43:52] you know, one particular solution doesn't fit all right,
[43:56] it's not one size fit all. You have to configure it every
[43:58] even if like people having Epic, right, Epic implementations and
[44:02] health plans, health providers, every single implementation of
[44:05] Epic is different. I mean, we worked with the
[44:08] hospital system. Every single instance of Epic
[44:10] they have across the system has got different data structures,
[44:13] right? So you have to Co create with
[44:16] the customer to adapt and configure it in stuff.
[44:18] So you build the infrastructure, you build the library of agents
[44:21] that are configurable highly. So we and then we deploy the
[44:24] forward deployed engineers to work with the customers and Co
[44:27] created for what they want to do, right.
[44:29] That's been our modus authority. So for what we do is we have a
[44:31] very robust platform road map that actually has capabilities
[44:35] it's coming that affects everybody.
[44:37] They have customer specific road map for releases that are like,
[44:40] you know, basically the agents that they want.
[44:42] They've actually we we work very consultatively and build out a
[44:44] road map for the customers and then go do that.
[44:46] And then there is like one of things that we see like the
[44:49] prior auth for pharmacy and the prior auth for medical and DME
[44:53] and some of these things everybody needs it.
[44:56] So that is a lot more product license ready to configure.
[44:58] So we can go live in 30 days, 45 days, right, compared to a
[45:03] claims automation capability, which it's OK to go live in 90
[45:07] days instead of 30 days, right? So it's, it's a very, I would
[45:11] say, you know, multi product approach and a platform and a
[45:16] multi product approach that we do it and it's lengthy sales
[45:20] cycles. But you know, the one thing that
[45:22] customers are not OK with is once the sales cycle gets done,
[45:25] you get a deal done. In most industries, the deal is
[45:28] done with Yahoo. You can celebrate in healthcare.
[45:31] Now you got to implement it. You got to make get them value.
[45:34] You got to go make that happen. That's when the real, you know,
[45:37] magic begins. So what we've done very well is
[45:40] like we tell customers, I'm going to give you 3 to 5X Roy in
[45:43] six months or at best 12 months, right?
[45:47] How do you do that? Well, you take, you know, don't
[45:49] try to eat the elephant, break it down into chunks.
[45:52] And that helps with our infrastructure being very
[45:54] modular. So you go in and say, look, your
[45:56] entire pride of process, you want to solve everything, but
[45:59] your biggest problem is that intake right now.
[46:01] I can solve it in two weeks. Go live there.
[46:04] Then I'll get the next module in that'll actually go, you know,
[46:07] take over the the medical summarization.
[46:09] Then you go to the benefit check and eligibility.
[46:11] So you give them a path to get to value much quicker.
[46:15] That's our single biggest secret sauce that we win with customers
[46:18] because we don't try to shove down.
[46:20] And here's the entire, you know, behemoth, you know, take throw
[46:23] away your existing system and use mine.
[46:25] Never works in healthcare. Absolutely.
[46:27] And when you talk about ROI and proving it to customers, what
[46:31] are the specific metrics that you used show that ROI to
[46:34] customers as well as use that internally to track success?
[46:38] Yeah. So look for metrics for
[46:40] customers, our biggest driving metric is efficiency, right?
[46:44] So for example, that is reduced cycle time, less time spent on
[46:48] the same workflow, what the parts of the workflow that they
[46:51] would otherwise have to spend time.
[46:52] So we track a lot of that and we report out.
[46:54] So customer KP is it'll be in there is multiple elements of
[46:57] this, but one big thing is efficiency.
[47:01] Second is throughput, right? Can I now do more because that I
[47:04] can, because I have more efficient system to do it.
[47:07] Third is quality and quality is could be like no, am I doing it
[47:10] net new better like you know, and there's it's arguable,
[47:13] right? Like for example, we learned
[47:15] very quickly that I mean, it's a chicken and egg problem if
[47:18] you're doing adjudication. And if it's an approval, you
[47:22] don't need a human to touch it because you're saving time
[47:25] because they're anyway going to approve it at the end of the
[47:27] day. But you also are capturing
[47:29] insight on like what leads to approvals versus not.
[47:32] So that will allow you to go shape your policies a little
[47:35] better. So you can go, do you know
[47:37] medical policies better? You can do that if it's so
[47:39] efficiency cycle time, you know, so efficiency also means cost
[47:44] reduction. So we manage like how do you
[47:45] reduce the cost of running operations for you?
[47:48] Then quality and quality comes in multiple different ways. 1 is
[47:51] like, you know, if it's in health stars or you know, heat
[47:59] is or something it requires, it improves star rating for a
[48:01] health plan. If there's clinical quality,
[48:04] your improved health status, your population health, you
[48:07] know, metrics go up, quality goes up.
[48:09] So there's multiple different metrics that we actually tracked
[48:12] by agent and by copilot. And we roll it up to that
[48:14] particular implementation saying, look, every customer of
[48:17] ours gets a dashboard that says here's the 15 Co pilots that are
[48:22] running, here's how they're actually performing against like
[48:25] 1 is your classic utilization metrics, which is how many
[48:28] people are using it. What is a cycle time and stuff?
[48:31] And then how is that improved your cycle time?
[48:33] Like now I can make, you know, claims decisions in half the
[48:36] time that I used to do. I'm using, you know, 86,000 less
[48:42] touch points to do this, which translates to about two and a
[48:45] half, $1,000,000 a month of savings because you know, you're
[48:48] paying about $80.00 an hour for someone to do it.
[48:51] So we map it out from a business observability standpoint, but
[48:55] we, but the product actually pushes metrics in there.
[48:57] Internally. We have a few measures that we
[48:59] use. 1 is like, you know, that the general everybody in
[49:02] healthcare does it like how many lives do we cover?
[49:04] So we don't want to talk about it.
[49:06] It's not a bragging right, but it's basically like, can you
[49:08] impact enough people to, to with what you're doing?
[49:11] And we're, you know, we're very soon we'll be actually impacting
[49:15] about half of America's population with what we do,
[49:18] right? So that's huge.
[49:20] And then also from a velocity, we do a lot of things on
[49:23] measuring velocity, velocity of releases.
[49:25] How soon can we actually get value opening?
[49:28] A big thing for us is going to be time to value for our
[49:30] customers is a metric. We internally measure the saying
[49:33] like we said, we we told them 90 days to go live, can we go live
[49:36] in 45 and surprise them and give them a thing.
[49:39] So things like that that we work on.
[49:42] But yeah, it's a solidly about. Absolutely.
[49:45] Thanks for sharing that. Now to the last couple of
[49:48] questions before we try to wrap this up.
[49:50] Looking back, what was the most pivotal decision that changed
[49:53] automizes trajectory? Was it a multi agent approach
[49:56] focusing on operations or something else?
[49:59] You know, it's a good question. I never thought of that because,
[50:03] you know, it's never like honestly, one decision.
[50:07] It's often like a culmination of multiple decisions, right?
[50:11] So the biggest consequential decision is deciding who I want
[50:14] to build this company with. So my Co founder Chris, right?
[50:18] I mean, we are like two different people, but then come
[50:21] together we're like 1 + 1 = 616 kind of a force to reckon with
[50:26] and incredibly grateful for him to actually I had to heal him
[50:29] out of retirement and comfortable jobs to come and do
[50:32] this with me. So that was a huge consequential
[50:34] decision that we did. The second big decision I would
[50:37] say is like in the, you know, first our first few customers,
[50:40] right, We went with customers who we could build really deep
[50:44] relationships so we can understand and learn the
[50:46] industry as an early stage company.
[50:49] One of the biggest thing that most, there is a founder's
[50:52] regret that happens in most startups, right?
[50:55] Said, why did I end up doing this, right?
[50:57] Or why did I choose this one, right?
[50:59] Unless you have so high conviction.
[51:01] So there's a process that you cannot force, which is, you
[51:04] know, So what happens? And I, I see a lot of founders
[51:06] who off the gate, they have good pedigree.
[51:08] They've done very well. They go and raise like $30
[51:10] million in a seed round and then they can figure this out six
[51:14] months later. They're trying to squeeze the
[51:16] Jews out of it and say, I can't get myself to do anything
[51:19] because I don't like this. This is not what I wanted to do.
[51:22] And like there's all these things and they, it gets into
[51:25] first year. One of the biggest consequential
[51:27] decision we did was like, we wanted to go in, learn more.
[51:31] We only care about learning. We only care about learning the
[51:33] customer, learning how they think, how they operate and
[51:35] stuff and building really deep relationship because that gets
[51:39] you a, you know, humans, all outcomes in life was driven by
[51:43] humans and you know, people in relationships.
[51:46] And so we focused on that. So that was a big consequential
[51:48] decision choosing who to take our seed round and our we did a
[51:52] price St. round. Asset management ventures is our
[51:55] lead, you know, investor in that round, Skip Fleshman.
[51:58] Getting them to actually invest and having Skip on the board was
[52:02] a huge decision. Skip comes in with a lot of
[52:05] experience on, you know, in healthcare and company building
[52:10] 30 plus years and he's like the Yoda of company building.
[52:14] So really helped us navigate early decisions, early things.
[52:17] Hey, don't worry about it. You are worrying too much about
[52:19] this thing that is not a problem.
[52:21] Focus on this, right? That kind of stuff, governance,
[52:23] all of that stuff was a great friend to to help us navigate
[52:25] through it. I think, you know, the last few
[52:28] years, I think the biggest thing we did was actually doubling
[52:31] down and we knew very quickly. I think I would say the big
[52:33] consequential decision, one of the big ones is actually we used
[52:39] life sciences as the initial entry point for us in
[52:41] healthcare. But when we started seeing the
[52:44] traction on health plans and health enterprises, because what
[52:49] we learned was like the biggest problem in Healthcare is
[52:51] efficiency. It's the cost of healthcare and
[52:54] cost leads to the quality of care problems, costly to the
[52:57] access to care problems, everything.
[52:59] So that was the lever that we can unlock and who has the
[53:02] largest cost in doing that? It's healthcare and you know,
[53:06] not pharma. Farmer didn't cares about cost,
[53:08] but they care about drug prices and stuff too.
[53:10] What these guys care about like health plans and which is
[53:14] insurance companies and health systems are the ones who care
[53:17] about cost. So that was, I wouldn't say a
[53:19] pivot, but because we are we did a lot of work that is palatable
[53:22] from farmer to here. But then just doubling down on
[53:25] our approach to go make that happen was a big consequential
[53:28] decision. I think this, you know, Series
[53:30] A, I'm hoping in a year from now, I'll look back and say this
[53:32] was a huge decision finding the right partners on our board, you
[53:36] know, the right approach for it. Cigna group ventures have been
[53:38] great. You know, we, we do a lot of
[53:40] work with the, the, the, the, the group and you know, this is
[53:44] great to have them also financially aligned with us and
[53:47] things too. So I think it's incredible.
[53:50] You know, it's a startups are always like, it's not one
[53:52] decision. It's like the microcosm of like
[53:55] a million decisions that you make on a daily basis.
[53:57] So that's the way we approach it.
[53:59] But great question though. Absolutely.
[54:03] To the last question before we try to wrap this up, for
[54:06] founders targeting heavily regulated industries like
[54:09] healthcare, what's your playbook for building credibility and
[54:12] navigating compliance from day one?
[54:16] You know, it's a good question. Well, compliance is one thing.
[54:19] Compliance you can navigate. It's not as hard as people make
[54:22] it to be. It's just stricter.
[54:24] You know, you just can't be loose.
[54:25] You can't, you got to have, I got to have AI can't just have a
[54:28] open door. I have to have a badge at the
[54:30] door. I mean things like that which
[54:31] you can figure it out. That's easy.
[54:33] I think the harder part is like getting into a regulated
[54:35] industry. It's the might, it's a, it's a,
[54:38] you know, I like to say something to my teams all the
[54:41] time and it's no, it's very true.
[54:44] It's actually from the Bhagavad Gita, right?
[54:46] Which is like, give up the fruits of the action.
[54:49] So I like to say like impatience for action and patience for
[54:54] results. That's what I tell everybody,
[54:56] right? Healthcare or any regulated
[54:58] industry, that's what you need to do.
[54:59] You have to have the impatience because otherwise you're not
[55:01] going to make any changes. You have to be, you know,
[55:04] urgent. You have to treat this problems
[55:06] are like we had to solve it right now and stuff like that.
[55:09] And then you have to have the patience for the results to come
[55:11] up. It's going to take a longer
[55:12] time, longer sales cycles. It's a lot of mind shift
[55:16] changes. People have done operations in a
[55:18] certain way. So that's one big thing.
[55:20] So my is a founder starting there.
[55:21] Usually like you should think about it as like, you know, read
[55:24] the Bhagavad Gita. It'll help you actually reframe.
[55:26] You know, karma yoga will give you a framework to think through
[55:28] it, right? So that's one impatience for
[55:30] action, patience for results. The other big thing I would say
[55:32] is like, look, every industry warrants change and no other
[55:37] industry than a regulated industry needs more change,
[55:39] right? So there is find the Champions,
[55:42] go find the folks who want to change and then listen deeply to
[55:46] them. Just embed yourself, understand
[55:48] it. And because you learn a lot
[55:51] because what Chad GPD tells you or what you can do on a Google
[55:56] AI mode search or otherwise is not what this industry is all
[56:00] about. There are so many contradictions
[56:02] here. I mean, podcasts are great,
[56:04] blogs are great, people talking about this is great and stuff.
[56:08] But you cannot replace the one-on-one individual
[56:12] conversations with people to understand how they operate, how
[56:15] they, what they struggle with, what they leave what, what
[56:17] they're, what they're really excited about and not excited
[56:20] about Trump's everything else. So listen deeply.
[56:24] Find the change makers who want to change.
[56:27] You will find the detractors and listen deeply to them too,
[56:29] right? So I think those two things I
[56:30] would say are like some good thing.
[56:32] And then regulation is regulation, which just means
[56:34] you're paying more for lawyers you're paying more for.
[56:36] So you got to budget that in and then, you know, you find the
[56:39] right people to do that for you. So that's the way I would see
[56:42] it. Absolutely.
[56:44] Yeah, that's one beautiful learning from Bhagavad Gita and
[56:47] one of the main ones if you want to take one from the entire
[56:50] book. There are many, of course, but
[56:52] thanks so much for doing this with us, Ganesh.
[56:55] It's been an incredible conversation.
[56:57] A lot of insights for me as well that I can apply to my
[57:00] businesses as well, and I'm sure for our listeners as well.
[57:03] Thanks so much for doing this. Do let our listeners know where
[57:06] they can learn more about what you are doing at Autonomize AI
[57:09] and where they can follow you. Yeah, no, thank you.
[57:12] Thank you, Rashan. So, yeah, you can, you know,
[57:14] reach out to us, look at follow us on LinkedIn Autonomize AI or
[57:18] autonomize, autonomize dot AII. Also I mentioned to you, I also
[57:22] hold that I I run a podcast called Stories in AI.
[57:26] We can, you know, if you want to follow that, it's on Spotify and
[57:29] all the platforms. What I've done there is actually
[57:32] try to get folks who are deep in the industry, not just
[57:36] healthcare, but other industry in getting their perspective on
[57:39] what how they think about applying AI to their lives.
[57:41] And that could be CD OS and CI OS thinking about how do you
[57:44] deploy, how do you manage it or business users for like, I'm
[57:47] struggling with this and I'm going to do this thing.
[57:49] So and I do believe that AI is such a critical piece of
[57:54] humanities history as we go look back from 100 years from now
[57:58] that you want everybody to take part in the build process of it
[58:02] than just be affected by it, right.
[58:03] So thanks so much for having me and it was such a pleasure and
[58:07] thanks for the great questions. Absolutely wish you all the
[58:10] best. Thank you.

Shows are listed from Apple Podcasts' charts and directory, and Podcast Index where it is used. Every show and episode belongs to its publisher and plays from the publisher's own feed; this site hosts none of them.