He bet on voice AI when no one else did. Now he has $1M+ customers & $100M raised. | Ankit Jain, Founder of Infinitus

Ankit left his job as a VC to launch a Voice AI platform—back in 2018! It wasn't the voice AI of today. The first demo sounded like a robot. But still, he convinced large enterprise customers in the healthcare space to try it out. He found a highly manual, call intensive workflow in the back office and autoamted it using AI.

Years later, he's raised $102M and has dozens of large $1M+ enterprise customers using the product. He talks about how he started it, how he saw the AI opportunity so early on, and how he found a way to lock in champions that pushed his product inside large enterprises.

Why you should listen:

• Why the search for product-market fit never stops.
• How to build trust with enterprise customers.
• How to get champions to fight battles for you and win enterprise deals.
• Why even successful startups are never straight lines up and to the right.

Keywords
product-market fit, AI in healthcare, automation, conversational AI, startup challenges, scaling, founder advice, technology evolution, compliance, trust in AI

Timestamps
(00:00:00) Intro
(00:02:01) How it all started
(00:06:29) Automating Insurance Calls
(00:16:37) Landing the first customers
(00:22:02) Giving your persona phone number to early users
(00:27:43) Landing large enterprise customers
(00:33:45) Getting Early Adopters to Believe
(00:36:34) Finding Product Market Fit
(00:38:04) One Piece of Advice

Chapters

• 2:01 How it all started
• 6:29 Automating Insurance Calls
• 16:37 Landing the first customers
• 22:02 Giving your persona phone number to early users
• 27:43 Landing large enterprise customers
• 33:45 Getting Early Adopters to Believe
• 36:34 Finding Product Market Fit
• 38:04 One Piece of Advice

Transcript

Ankit Jain (00:00):

Every year, there's a moment when you feel like you found product market fit for that stage, and then you kind of work your way into that, and then you hit some kind of a ceiling, whether it's in your mind or from a business perspective. And you have to work hard to kind of punch through that and make the ceiling your next floor and grow from there. So I don't think it's a, it's a, once you find it, you've found it, it's a constant iteration and growing your ambition, growing your vision, growing the impact you want to have. There's days that are exhilarating and there's days that just feel like, why the hell am I doing this? , right. And , and you know, like there's many such moments in any startup's journey. , you know, there's days when you sign your biggest deal and someone is like, you know, a Fortune 10 company goes, we're going to use you and scale this up to, , millions of calls a year. And you go, this is incredible. , and then there's days when a big insurance company that you call all the time suddenly says, we are no longer taking AI phone calls. And you're like, all right.

Pablo Srugo (01:00):

So I open up listennotes.com today, which is a platform that ranks podcasts worldwide that I've often used to understand kind of where we're at. And as of today, we are now a top 1% podcast globally that's across all the millions of podcasts that are out there. So I just wanted to let you know this because we're building this community together, and I wanted to say thank you, thank you for listening. Thank you for following the show. Thank you for sharing it with other founders. I love the fact that I get to do this. I love that we continue to grow, and I love to end the year on such a high note. Let's do it again next year. Here's to 2025. Welcome to the show, man.

Ankit Jain (01:38):

Thanks for having me, Pablo. It's gonna be fun, dude. 

Pablo Srugo (01:41):

So you just raised just over $50 million from Andreessen, you know, one of the most well-known, , VCs in the world. And you're doing like, you know, AI agents in healthcare, which I think these days everybody's doing AI agents, except you've been doing this since like 2018, 2019, you know, five, six years ago. How did you kind of get into this world so early on

Ankit Jain (02:01):

Before that, I helped start Google's AI venture fund Gradient Ventures, right after the attention is all you need paper came out. So the thesis there was, the world is gonna change because of these breakthroughs across every major industry. And how can Google invest in that growth, both from its own plug platform, but also innovations around the ecosystem. And so I had a front row seat to see how this technology was being used in many different ways. , and one time I was showing my wife a demo of a, a project that was automating phone calls to make spa and restaurant reservations. And she said, Ankit, incredible technology. Good demo, but why do you need this to make spa reservations? I wish someone would bring this to healthcare. She spent her career in healthcare and she said, if you could make information and knowledge available to people when they need it, how they need it, if you could unlock time that was scarce in healthcare where data is exchanging hands on phone calls, it would change how healthcare in the US works.

Pablo Srugo (03:01):

And this was what- the demo. Like I kind of have a hazy kind of recollection. This is something that Google actually showed the world right back in the day? 

Ankit Jain (03:08):

Yeah. This was the 2018 Google io Google Duplex demo, where you could say, Hey, Google, make me a reservation at a salon. And she said, great technology. What a terrible use case.

Pablo Srugo (03:19):

<laugh>. Well, it's also crazy that, you know, that that thing was demoed so long ago. And I still struggle to talk to-  when I say, Hey, Google <laugh> like my, my Google Home speaker, and you know, it took like chatgpt to really bring a few of these, like, to get people thinking about this stuff again, even though Google was there kind of so early.

Ankit Jain (03:37):

Yeah. You know, I think a lot of it comes down to the conversational ui, if you will. If you go after things that are extremely narrow, then people don't have their imaginations working on overdrive, right? I think when Google said we can automate reservations, it didn't spur everyone to go, how do I go apply this everywhere in the world? But when chat GBT opened up an interface that you could talk to about anything, people started creating art, creating poems, creating reports. You know, there's this project called chat, PRD which is automating what a product manager should do in writing A PRD. It asks you questions, it has a conversational UI to it. And that idea of a conversational ui, I think was really powerful in getting all of us to dream bigger.

Pablo Srugo (04:27):

And so what do you do when your wife says that to you, like, do you just kind of quit right away? Like, is it, is it that strong of a signal that just pulls you out? Or do you you kind of ruminate on that for a while?

Ankit Jain (04:40):

From the time I had that conversation with my wife, to me starting Infinitus was about five months. So it wasn't that long a time. Now my day job was the vc. Now, prior to this, I had started a company. I come from an engineering background, so I'm very much a builder. I sold that company. And so there was a part of me that even though I was investing in entrepreneurs and being a partner to them and enjoying that, I was really missing building a company. It was fun to build a venture capital firm, but I was ready to kind of build a company again. And so my co-founder and I were ruminating on a number of different ideas, but when this one came, there was something about it that pulled us towards it because I think healthcare is mostly worked on by people that get passionate about the industry.

Ankit Jain (05:27):

It's something about improving it, not just for somebody else. You're not building the next ad system, the next gaming platform. You're building something that's gonna serve you and your loved ones as well. And so we started digging in deeper. I didn't know much about healthcare. I'd done a couple of investments in the space, but I am very much a newbie. So I started doing diligence calls by calling prospective customers and saying, Hey, I'm a VC at Google. There's this company that I'm looking at that says they can automate phone calls in the back office. Does that sound interesting? And every single person I reached out to said, that's a huge problem. The first one said, we have 3000 people doing it. The second one said 400 people. The third one said a thousand people. So we had validated it and they all said, how can I meet this company? And I'm like, I'll connect you with them in a couple of months. And, you know, lo and behold, a few months later I said, oh, that's my new company called Infinitus. What,

Pablo Srugo (06:20):

Like what kind of calls? Were they talking when they say they have thousands of these sorts of people or whatever calls are happening? Like what are these calls about?

Ankit Jain (06:28):

So one of the things that is hard in the US system is that a doctor can't independently make a decision on what the next treatment for a patient should be. There's checks and balances in place. 'cause Someone's gotta pay for it. It's the insurance company that pays for it. And so for the high cost medications for patients with chronic diseases doctors have to get what's called a prior authorization. And sometimes they need it, sometimes they don't need it. So in the back office, there's clinical, there's administrative staff that takes an order from a doctor that says, this patient has a macular degeneration and therefore needs a medication called Eylea, which is tens of thousands of dollars. They call the insurance company to say, is this covered? That's question one. If it's covered, do I need to prove anything before you'll pay for it?

Ankit Jain (07:17):

What is the patient's part of what will be, what does the patient have to pay and what will you pay? So there could be a co-payment, a flat fee, $20, $25, or a co-insurance, in which case a patient is responsible for 20% of the total cost. So they do that research, it's called a benefits investigation before they then initiate a prior authorization and then check on the status of the prior authorization. So all of these phone calls that are happening, and that's what these big groups were doing for thousands of doctors around the country. And they said, if you can automate it, it would reduce the time it takes to get a patient on therapy. It would reduce the administrative cost in the back office, which through trickle down theory would reduce the cost of healthcare in the us. And so there's a lot of second and third order benefits there, but the primary benefit is reducing that time to therapy for patients.

Pablo Srugo (08:06):

So you're automating not calls coming into like any sort of healthcare, let's say hospital or whatever you're talking about, kind of admin calls going out to insurance companies that you're automating first?

Ankit Jain (08:15):

That's where we started. And you know, over the last few years we've expanded that surface area quite a bit. So we started by calling insurance companies. Now the insurance companies said, well, why the heck are my humans talking to your machines all day long? And we said, they don't have to gimme the data digitally and I'll get rid of the phone calls. It's a win for me. 'cause That's instantaneous and it's a win for you. 'cause You don't need to have a person answering questions to us or others. And some of the biggest ones now we work with where they give us as much of the data as they have digitally through APIs, it's still not a hundred percent, it's 50 to 70% digitally. So it reduces that 30 minute call into a 10 to 15 minute call, which is a saving for them and a reducing the turnaround time for us.

Ankit Jain (08:58):

Eventually we wanna make that back office as instantaneous as possible. Now as we started doing this, we were serving providers, but we started serving a lot of pharmacies that also have to do the same interaction with the peers, the insurance companies. And then they said, Hey, can you also start calling the co-pay assistance programs? Because remember, there's some patients who are responsible for 20 to 30% of the price of these expensive drugs and they can't afford that. So then they said, can you call these foundations that cover patients that don't have the means to pay for it and see if there is money left in the foundation? Is there balance for this patient? Again, that's phone calls that were being done by their staff, but it was often delayed. This allowed it to happen faster, quicker, and better. In the last year and a half, we've started seeing an increasing demand for patient facing phone calls to say, Hey, we know you're about to go on your, for your first infusion. Do you have any questions about what that experience is gonna be like, educating them on the journey that they're on, educating them about their diagnosis and just being available to help them through what's one of the hardest journeys in their life. 

Pablo Srugo (10:08):

And so, let's go back to that kind of early day. Like, so I understand obviously it expands, like as you start automating one phone call, it just, you know, and people see the possibilities, it's a bit of a no-brainer to start using it in more places. But like at the beginning, you know, the thing that's actually in the back of my mind, just for a little bit at least, is just like the tech side is, you know, I understand that Google demoed this thing, but there's a demo and then there's production. And like, you know, for an average user voice AI didn't go into production until like a year or so ago. What was the tech like in 2018, 2019, 2020? Like, were you able to actually launch a product back then? What were you using to build on top of? 

Ankit Jain (10:43):

I will say voice AI is really coming of age now. I think when we compare ourselves to some of the folks that are starting companies now, they're like, Hey, we've done a hundred thousand calls. We're like, we're doing millions and millions of these calls. And you know, we started doing them in production in late 2019. So we started the company in February, 2019. About six months later we started doing our first calls in production. And you know, the LLMs of that time weren't nearly as capable as the ones we have today. And forget LLMs, I mean, in retrospect they're pretty small language models. They were knowledgeable. They were able to be used for things like intent detection, which is kind of step one, when you break down the technology, there's three things that are happening the traditional way.

Ankit Jain (11:29):

A conversational AI system is built. You hear what's coming in on the audio, you take the speech and you make a text. Then you decide it, you do natural language processing on the text to understand the context of conversation, what's happened so far, this latest utterance, what was said. And then you determine what action to take next. Are you gonna say something back? Are you gonna answer a question or are you gonna ask a question? Are you gonna clarify something or are you gonna stay silent? 'cause If someone says, can you hold on for a minute? You say, sure. And you wait silently until they say something back and then you take that next action and you generate voice back. Right? And over the last five, six years, every piece of that has improved. You've improved your speech to text, you've improved your NLP with better LLMs and you've improved the voice generation, the voice synthesis pretty dramatically as well. And you realize that you lose a lot of granularity when you go from voice down to text. You lose emotion. You know, you ask a question of somebody, Hey what's your name? And you say, Pablo. And I say, okay, that's fine. But if I say hey, what's your name? And you say, okay, it's Pablo. And I say, what's your name? And I repeat it three times for some reason it's not working. And then you say, okay, that sounds different. Like that voice.

Pablo Srugo (12:45):

Of course there's a lot communicated above and beyond just the word 

Ankit Jain (12:49):

That's right, yes. There's a lot more signal. Right? And, and, and so we did a lot of research because it turns out there wasn't much research on combining audio and text models that was out there. 'cause Everyone was doing NLP when it came to text. So some of the publications we have at some of the big conversational AI conferences and just NLP conferences are about combining text and audio models and how to use them to have a better conversational UI. And all of those things we had to build the guardrails we had to build. 'cause Again, the hallucination rates were significantly worse in 2019 than they are today. You know, I internally joke, I think the hallucination rates have gone down, but the ones that happen are so hard to detect that they can, you can get away with them. So the guardrails we built in 2019 and 2020 are serving us as well today as they were then because we had built it for a way worse system that now it gives the right guardrails and gives our customers a lot of confidence that what we're doing is not gonna go off script in a way that will cause compliance issues, which is a big deal in healthcare.

Pablo Srugo (13:55):

How did you get something into production? You just constrain it a lot. Was your use case very hyper defined? Did you have to build the full stack or was it like assembly or something else, like around back then that you could leverage?

Ankit Jain (14:07):

No, we've built most of the stack ourselves. A lot of what you see today are generalized versions of that. But when we think about the phone calls we are doing, the average phone call we automate is over 30 minutes long. And so we've been doing this for five years. And these are highly technical in nature with all kinds of standard operating procedures. And you know, I'll give you one example. When we call an insurance company to get benefits for a patient, 25% of the time we get incorrect information. So you can't- you ask, what's the copay for this patient, for this drug? You can't just trust what's being said because that person is opening up A PDF flipping through it and reading it. What if they open the wrong PDF? What if they open the wrong page?

Ankit Jain (14:53):

So in real time we're doing QA on it, seeing all the data that we know and saying, are you sure? Can you check that please? And that level of complexity, you can't take something off the shelf. And you really have to tune it to healthcare, tune it to the kinds of use cases we're doing both on the administrative and on the clinical side. And you know, as we've done the patient facing calls, knowing how drugs interact with each other, understanding the kinds of adverse events that happen, tying all of that together requires a lot of custom model building. 

Pablo Srugo (15:23):

So, you build something custom back then. Do you then, do you hyperfocus it? Like do you have, do you have like a very specific, 'cause a 30 minute call can go in so many directions, I can't imagine back then being able to handle, do, do you have a human in the loop? Like I'm just trying to, I'm just trying to think like how do you, you know, bring what's there to actually bring some value given all the complexities?

Ankit Jain (15:41):

Of course. We, you know, we took a lot of inspiration from our friends over at Waymo. I think one of the things Waymo did very well in the self-driving world was they said, let's get a car on the streets as early as possible. And if you need to have a safety driver, that's okay. That safety driver when they take over gives you training data to improve your system. And as you do millions of miles of driving, you get enough for the machine to start determining its own next specs action. And now you've got way more cars in cities around the world driving themselves. And that was, you know, their journey took 12, 13 years. Our journey is much more compressed 'cause we are starting from a better place from a technology perspective in 2019. So we had the ability for the machine to raise its virtual hand and say, I need help. And someone could help the machine kind of take its next steps. In the beginning when we started it was machine driven and now it's mostly, you know, it's much more automated than it was then. 

Pablo Srugo (16:37):

Did you raise money like right outta the gate?

Ankit Jain (16:39)

We did. 

Pablo Srugo (16:40)

What was like your first round and like, what was the timing of

Ankit Jain (16:42):

Yeah, so in 2019 we raised just under four and a half million from Kleiner Perkins and a number of angels.

Pablo Srugo (16:49):

 And who was your first customer and how'd you get them?

Ankit Jain (16:52):

Our first customer is a company called Syncora. They're a big supply chain company like McKesson and Cardinal Health, the three big supply chain companies in healthcare. And they support every major pharmaceutical manufacturer, every major health system in the country to run many of their back offices. They're a Fortune 10 company. And they were one of the people that I had sent a cold email to and said, I have this company that I'm diligencing. And they said, we've got thousands of people. And we worked with them. It took a lot to get them going. 'cause Again, this was, they, everybody in 2019 thought we were doing black magic. No one believed a machine could have these kind of phone calls. And so we, you know, went through all the trials, tribulations, set ourselves up with all the security and compliance again.

Ankit Jain (17:39):

A Fortune 10 company isn't going to just let personal health information into the hands of a no-name startup with no other customers but kind of worked very closely hand in hand with them to get the system up running and live. And you know, they've spoken about this publicly, but the way it's scaled for them really helped them run their operations in a more efficient way. But more than anything, it reduced that turnaround time and it increased their quality. I think they said it was 10% more accurate on average than their teams, which is pretty amazing.

Pablo Srugo (18:10):

How and how did you structure that? Like, did they give you like, okay, you're gonna take over 5% of the calls or some specific set of the calls, like how did you go from, you know, whatever started to, you know, fully deployed?

Ankit Jain (18:21):

everybody starts in a limited way and then scales it up over time. 'cause You wanna try it out, you wanna do side-by-sides and over time you build that trust in the system and you scale it up.

Pablo Srugo (18:32):

I guess I'm curious about those early days, like any stores you might have around trying to get this out. Like, you know, I just feel like hard tech comes up against all sort of, you know, bumps, bumps along the way from dream to reality

Ankit Jain (18:45):

I mean, I'll tell you a couple of fun stories. So before we started the company, my co-founder, Sham and I, so Sham and I have known each other since middle school, went to middle school, high school together, and college roommates, we did our last startup together. So when I told him about this conversation with my wife, he said, cool idea. Lemme try to build a prototype. And I said, you know, let's build as lightweight a prototype as possible and let's make a phone call and see whether the other person even talks to it. 'cause The people on the insurance side won't talk to it. There is no business. So he built a very lightweight prototype and we called United Healthcare. Our friends at United know this story, so that's why I feel comfortable kinda sharing it. But we called, got through the IVR and got to an agent and the agent said, can you tell me the patient's name and date of birth? And again, this was like Wizard of Oz. We were typing behind the scenes, right? 'cause We're, we didn't have a fully automated system.

Pablo Srugo (19:35):

You're typing the text and you're just testing the voice. The voice part of it. Yes.

Ankit Jain (19:38):

And we just wanted to see whether the person on the other side will talk to, would

Pablo Srugo (19:42):

Believe the voice or is it too robotic? Yeah. Okay.

Ankit Jain (19:44):

Well no, it was a hundred percent robotic whether they'd be comfortable talking to a machine. 

Pablo Srugo (19:50):

Right. Oh, knowing- so this isn't like the today’s stuff where you're like, is it a human? Is it a machine? It's like, is for sure a machine, it's just are you comfortable talking to it?

Ankit Jain (19:56):

Exactly. right. Or are they gonna just hang up? Right, because there weren't any realistic voices available back then. We call in and we're like, Hey, you're looking for a patient. The benefits for a patient, what's the patient's name at the moment? I'm like, Bruce Willis <laugh>, what's the patient's date of birth? January 1st, 1947. something like that. Made it up. And then they were like, let me look that up. And we're like, looking at each other. Oh my God. Like they're actually talking to us about it and they're like, sorry we can't find this patient. We're like, well yeah, 'cause we made that up. We're like, oh, we're sorry. We'll check the information and call back again. Thanks so much. And we hang up and we're like, that was the moment of cool. There's something real here right now. That was one call.

Ankit Jain (20:38):

You know, there was a couple else we did where they hung up on us and we're like, all right, so there's some hope we need to work through this. And you know, there was someone we were talking to in the early days and we were like, Hey, we get a hangup rate back then, like 30%- people aren't talking to us. And someone said, that's pretty amazing because this was a friend of mine that worked at Uber in the early days. He's like, every city we went to wanted to ban us. You, you're starting from a place where you have 60 to 70% acceptance. We started with 0% acceptance and worked our way to get approval, approval, approval. And so you're starting in a better place. That's incredible. You're trying to change the world to go from human to human communication to machine to human and eventually machine to machine. And that's instantaneous APIs, right? And so he's like, you're starting in a place of strength. You should never think about the 30 or 40% that get hung up on. You should think about the fact that 60% of the time you've moved into the future. And that was always part of our inspiration as we made progress on that front. 

Pablo Srugo (21:36):

Well, I would actually, I would agree with him. You know, like 60, 70% talking to a machine is very high. I'm wondering like now looking back or even at the time, like how do you explain that? Like what is it about that use case? Because frankly, most people just hang up, hang up on machines. Like even today, like you hear a machine like robocall hang up, spam, whatever, right? Like all these sorts of things come into your head. What is it about this use case that got you, you know, 60, 70% of people to, to be willing to, to give it information? 

Ankit Jain (22:02):

So a couple of things. I think context matters a lot when you get a phone call on your phone from a number you don't know, most of the time you won't pick it up. Sometimes you'll pick it up and figure out who it is and then hang up quickly. But if you get a call on your phone which says CVS or your local pharmacy, you're more likely to pick it up because you go, I know what that brand is. And if you're getting that phone call now from a known brand, when you're expecting them to call you because a doctor sent a prescription in, then you're really likely to pick it up, right? So knowing the context and having some trust helps. So when you're calling an insurance company, what's the context there? One, people don't just go through and wait on hold for half an hour for fun.

Ankit Jain (22:45):

So by the time you get to an agent, you've authenticated yourself in the IVR, you've given all kinds of identifying information, you've waited on hold. And then when the patient, a person picks up, you again authenticate yourself and ask very relevant questions. So that team is not getting a lot of spam calls 'cause spammers don't have half an hour to wait to get there. And they don't have the identifying information. And so there were certain groups that used to ask us- another fun story from the early days. Anytime you call an insurance company just like when you call any customer support number, one of the first questions they ask you is what's the best callback number in case we get disconnected? Well, for the first three years, that was my cell phone, right? And so every time the call got disconnected or something was incorrect, I used to get a phone call.

Ankit Jain (23:29):

And once in a while that phone call was, Hey, I am a supervisor of a call center team here <at the name of a big insurance company in the world>. Who the hell are you and why are your robots calling my team? And I was like, oh, nice to meet you John. My name is Ankit and I'm the CEO of a company called Infinitus. And we automate benefit verifications and prior authorizations for tens of thousands of doctors and providers around the country. And I appreciate you calling me. And that was the startup, what we now call peer relations, educating them on what this can do, how it can drive efficiency for them. And they go, oh, this is kind of cool. I'm like, by the way, I know you called and you were confused, but what did your call center agent think about the experience?

Ankit Jain (24:12):

He is like, I hate to say this, but they kind of liked it. And I was like, why? They're like, because you weren't yelling at them. The average person who's waited for half an hour is impatient, wants to get through it. And sometimes their mic is too close to their mouth or too far. So it's hard to understand. They're inefficient. Your digital assistant, your AI agent is really efficient, is really easy to understand and can answer questions, and can have a full conversation. I'm like, that's interesting. So it also helped us build that value proposition for the people receiving our phone calls. And they're like, oh, by the way, we love the fact that you call from your phone number, the same phone number all the time. You have one voice. All of our customer service reps know your phone number, know your voice. And so they feel that trust with it over time. So that's been built over the last four or five years. 

Pablo Srugo (24:56):

Did you ever close a sale or something like that off somebody calling your phone number? Because they were like, <laugh>, they're upset, but then you turn, you turn it around?

Ankit Jain (25:03):

it's happened multiple times. Many of the biggest insurance companies in the country, in addition to the doctor side, are now using us because they first heard about us when they got a call and now they're like, we make millions of calls as well. Normally they're backlogged, but we can use you to work through that backlog and get into the future.

Pablo Srugo (25:22):

You love this show, you don't wanna miss the next episode. Why would you? so hit that follow button? Trust me, it's in your own best interest. It's funny, I think you'd be surprised, maybe not, but like I certainly was surprised by just how many founders in the early days that we had on this show that became, you know, very successful, gave up their phone numbers to users and, and customers. Like in the early days for like probably way too long. Like I had somebody give like a 1800 number, you know what I mean? Like he's getting calls 24 hours a day. I had somebody else give their phone number as the founder of WealthSimple. it's like the Robin Hood of Canada. I mean you talk about like, you know, users like just not, not like big enterprise, right? Like anybody can call your cell. But I think there's something about that, you know, just being willing to stay so close to your customers you know, in a bit of an insane way, like having no boundaries on purpose by design that I think can be extremely helpful. 

Ankit Jain (26:12):

Yeah, I mean that's how you have a pulse on what's actually going on. You know, I I think Parker Conrad at Rippling still runs their internal HR and IT solo administrator. 'cause He is like, if this is what we're selling to everybody, I wanna know how it works and how it doesn't work. And can I run a 5,000 person company solo from an HR and IT perspective? 'cause If I can't, then others won't be able to do it with 10 people. And so you know, I think founders often do put themselves in these places with no boundaries. And I think at least for us, both sham and I do it in different ways and it helps us keep a pulse of how we're improving things over time.

Pablo Srugo (26:48):

So, you know, walk me through this, there's some value props that are obvious immediately. And frankly, like the entire wave of AI agents today is like that. And that's why there's so much money being thrown at in so much competition. And I think that the big problem today is trying to figure out who's gonna win which category. Like where is there gonna be too much competition? 'cause The use cases are no-brainers. Like if you can automate a routine phone call in your case and turn it from something that costs you, you know, a human to something that costs you a machine price and all the other benefits that you, that you named, it's a no brainer what. But then again, like the flip side is nothing's that easy <laugh>, you know what I mean? Like, nothing in startup land is easy up into the right. When you think back to those early days, what were the hard parts of this? Because I assume it wasn't really like getting people, you know, for the most part. I don't know, you tell me like was it getting people to try it out? Was it getting people to scale or was it just building the thing and getting it to work? Like what, where, where were the punches to the face kind of coming from?

Ankit Jain (27:43):

Yeah, there's all kinds of punches. and I think it's right to say that there's days that are exhilarating and there's days that just feel like, why the hell am I doing this? Right <laugh> and and you know, like there's many such moments in any startup's journey. You know, there's days when you sign your biggest deal and someone is like, you know, a Fortune 10 company goes, we're going to use you and scale this up to millions of calls a year. And you go, this is incredible. And then there's days when a big insurance company that you call all the time suddenly says, we are no longer taking AI phone calls. And you're like, all right, we've gotta work with them to figure out why did they block all AI phone calls. And turns out about a year ago some of the biggest ones blocked it.

Ankit Jain (28:29):

And that was because there was a handful of companies that started calling them and they were terrible. And they were like, we can't have different policies for different AI agents. We know you guys are good, we know you personally, but we had to have one policy. 'cause We're not gonna teach our call center employees around the world to know the difference between you and somebody else. To them it's machine or machine. So as we worked through that with some of them, then it led to another one of those exhilarating moments. Most of them now have dedicated call centers for our AI agent. We have a private phone number that we call which reduces our wait time. We get data digitally from them, and then we get a set of human agents who love talking to our AI agent and it goes really fast.

Ankit Jain (29:13):

So it's a win for them and it's a win for us. But again, that's one example of how that happens in the last two years because of the explosive growth of AI agents. Al almost every enterprise has put into place AI review boards or machine learning review boards to make sure that when they adopt a new AI agent or a new AI platform, they're not introducing compliance issues, biases. Again, these are, the intentions are extremely good. The problem is most enterprises aren't very AI savvy. And so they tried to put the same 10 check boxes in front of every vendor whether they apply or don't. And so a good example of this, we were talking to a large insurance company. They wanted to start using us to do back office phone calls of the kind I described between two different insurance companies. And they said, how have you tested your AI agents for biases?

Ankit Jain (30:09):

And we gave them a good answer, but how about how do you test for people that have English as a second language in the back office? And I said, well, folks, like when I call your call center, I try to get to the call as efficiently as possible because that's what you would want me to do. I don't start off with a survey of, before we can get started, can you tell me your age? Can you tell me your gender? Can you tell me whether English is your first or second language in order to then figure that out. I'm just calling a call center and whoever picks up that's the person I talk to and because it's an English call center, they can understand me and they can, they can talk to our AI agents. But it took a good amount of back and forth to get them over that to go, oh, this is not the same as a patient facing use case where you do wanna make sure that you test for equity and across all these different verticals and different axes and that they shouldn't use the same check boxes that they do for patient facing AI agents as they do in back office AI agents.

Ankit Jain (31:09):

But we're in that journey because it's early in this ecosystem to help the ecosystem mature. But that's, you know, that's another example of something you're like, you thought the deal was gonna close, but it took an extra three months or four months. As you're just working through these kind of challenges, 

Pablo Srugo (31:23):

there's always, I mean, there's always, there's always the unpredictable stuff. What about like, in the early days, like you think that through the early days, like what was the hard part of this? Which, which were the parts that were maybe more existential? Because frankly, like every startup, I, I mean there's, there's a handful, like, but even if you look at the super fast growing ones, like whether Snapchat or Facebook or whatever, then it's like, is somebody else gonna eat our lunch? Are we just gonna implode due to like, you know, chaos within, within the company itself? Like what about for you? Like what was, what was the hard part of getting this off the ground?

Ankit Jain (31:56):

Yeah, again, as I was saying in 2019, 2020 people thought we were doing black magic. And so just getting people to believe that what we were saying we could deliver on and again, we are focused on large healthcare enterprises. It's a chicken and egg problem where everyone wants to know who else is using it. And healthcare also tends to be one of those industries where people like security don't want to talk about their vendors publicly. And so you have to play all kinds of games to get your first five or 10 customers and then the industry suddenly hears about it everywhere and then it snowballs into something much bigger. But those early sales were a lot of effort, a lot of just finding the right champion that wanted to be part of the future and really before we could cross the chasm in the traditional sense. But like those early adopters, I'm forever grateful for them for having the vision and the courage to say, let's try this new thing on. I think the thing that changed for us was after Chat GPT, everyone goes, oh, that's all you can do. Why can't you also do this? why do I still need nurses? It's like, no, we still need nurses, we still need doctors, but some of the stuff we should have machines do and the other stuff we allow everyone to practice at the top of their license.

Pablo Srugo (33:10):

so I totally agree. I think with enterprise it's always- 'cause in a sense, like you kind of, you come into it and you have no credibility. Like as much as you have this promise of ROI, you know, nobody really believes you because you really haven't done it and the risk is so high and the upside for the like for the person is kind of low. Like they take a big risk on you. It goes out, maybe they get fired, they take a risk on you, it goes up like maybe they get promoted or maybe nothing happens. You know what I mean? Like, it's not clear ROIfor the human, even though it might be for the organization. How did you do that? Like how did you get those first few champions, those first few, early adopters to believe?

Ankit Jain (33:45):

I will say even to this day, every champion of ours has gotten one, if not two promotions and within the first two years of using us, like,

Pablo Srugo (33:53)

 there you go. 

Ankit Jain (33:54)

It's fascinating. It's actually one of those metrics that we track internally. Because you're right, the human part of it really matters. And if someone is going to put so much at stake to adopt something ahead of the curve, we wanna make sure we put them in the best light as well because they're doing a lot to take us there. Anytime you see a quote from somebody that talks about infinitus, that person has already had one, if not two promotions, which is pretty amazing. Like, I'm happy for them, but I'm also happy for our go-to market team and very proud of them, that they could identify that person who was hungry, wanting to make change, and then got rewarded by their institution for taking that risk. 'cause There're many examples of where that doesn't happen and where someone gets fired for taking a risk and it goes south.

Pablo Srugo (34:36):

But that's smart. I mean, even that feels like you put a little bit of a different emphasis, like instead of just thinking about the ROI to the organization, like you're thinking about the person who's gotta bring you into that organization, the risk that she or, or he has to take and what the upside is for them.

Ankit Jain (34:51):

at the end of the day, it's all about people, right? Like, we wanna make this better for the patient by reducing turnaround times. We wanna make it easy for our champions to tell our story internally. We wanna make it easier for the IT teams to integrate our stuff. You wanna make it easy for everybody. it's hard enough as it is. If we make it harder and not give it on a platter, on a silver platter, then it's never gonna get done.

Pablo Srugo (35:15):

Give me a sense maybe just of then the growth, like you had that first one you did a few months, and then did it just go like, company wide? How quickly did you go from like one to 10 customers sort of thing 

Ankit Jain (35:25):

at a given customer, It normally takes a couple of years to really expand into its full maturity and you know, like our first customer has grown 10 x in four years and is continuing to grow. And it's amazing to see that at the same time we're now landing much bigger deals because the surface area of our offerings has grown quite a bit. The kinds of phone calls we automate, the fact that we're both in the front office and the back office. The fact that we can do all kinds of insights and analytics on top of our phone calls, it allows us to come in and say, we've got a platform that can really help supercharge your teams. And that's a really powerful value proposition to the buyer because they go, this allows me to truly transform our business and the age of ai.

Pablo Srugo (36:14):

And what, like, what kind of size customers are you landing? These are like six figures, seven figure deals.

Ankit Jain (36:18):

Yeah. Our average customer is over seven figures.

Pablo Srugo (36:21):

Perfect. Well listen, okay, well we're gonna stop it there. But I'll end with the two questions we always end on. The first question is, when did you feel like you'd found true product market fit?

Ankit Jain (36:34):

You know, it's a tricky question because I feel like every year there's a moment when you feel like you found product market fit for that stage, and then you kind of work your way into that and then you hit some kind of a ceiling, whether it's in your mind or from a business perspective, and you have to work hard to kind of punch through that and make the, that ceiling your next floor and grow from there. So I don't think it's a, it's a, once you find it, you've found it, it's a constant iteration and growing your ambition, growing your vision, growing the impact you want to have. But I feel like every year, year and a half, we have that moment of, alright, how do we take this further? And then we find that inspiration and grow, grow more. But the first time we felt that was probably about a year into the company where - because we had gone live with our first customer, it was working and we had the confidence to go to everybody else and go, this idea we had is real. And that, and, and I think so much incredible stuff happens when you have the confidence to do it with real proof behind you. And that was the first one. But we felt that at multiple points in our journey. 

Pablo Srugo (37:43):

At this point, you know, you've had an exit, you've done dozens of investments, you've been on the VC side, now you're back on the founder side, and raised over a hundred million dollars just in this venture loan. Like, when you take everything that you've seen and done, what, what do you find is like, you, you must talk to a lot like early stage founders. Like what do you find are some of the most common pieces of advice that you tend to give?

Ankit Jain (38:04):

So my first company, I started it because I was 27 and I was thinking, if I don't bet on myself now, when will I do it? And I started a company to start a company. And, you know, we were lucky that we were able to get a good exit there. I don't think we had a, you know, a product or a company that was revolutionary, but you know, we, we built something that somebody else valued more than I think most of the world did, which is great. But something that Sham and I promised ourselves was we'd only start another company if there was a true market pain point and it was a problem that we were passionate enough about that we wouldn't mind kind of grinding it out. And somehow that brought us into healthcare where nothing is easy and you have to grind it out. But it's a problem that we're very passionate about and it's a very big market that is hungry for technology to really transform it. And so when I talk to folks, I say, find something that you're passionate about in a big market, because even if your early hypothesis is wrong, the pivots and the turns and twists you do will keep you in a large market and it's a market that you're passionate about. Well,

Pablo Srugo (39:09):

Thank you so much for sharing the story, man. It's been great.

Ankit Jain (39:12):

Yeah, thanks for having me.

Pablo Srugo (39:13):

You remember the first person who told you about Bitcoin, the first person who told you about Uber? You want to be that person because being first is cool. So be a cool person and tell your founder friends, set it to them on WhatsApp, put it in a WhatsApp group, put it on a Slack channel. Let people know about the show. Let people know about this episode. Don't let somebody else beat you to the punch and share it with your founder friends first. Remember what Ricky Bobby said, if you ain't first, you're last.