Subscribe to AIMinds Newsletter
Stay up-to-date with the latest AI Apps and cutting-edge AI news.
Subscribe
Share this guide

About this episode

“Deepgram has been a crucial piece of the solution. The problem is clinicians have to write a lot of documentation at the end of the day. And the solution for that is we have a AI scribe where clinicians can use them either to record the conversation between them and patient or a quick dictation. Deepgram has been instrumental to turn that audio into high quality transcription. The quality is so good that our user keep telling us, 'Hey, the program is amazing.' I was intentionally trying to mess with it by speaking to the phone next to a music player and they still recognize everything.”

— Yansen Zhou

Yansen Zhou is the Founder of Hippo Scribe, an AI Scribe product that helps clinicians finish documentation in minutes by generating a first draft from audio recording. Previously, he was a software engineer.

Listen to the episode on Spotify, Apple Podcast, Podcast addicts, Castbox. You can also watch this episode on YouTube.

In this episode of AIMinds, Demetrios and Yansen speaksabout the challenges and triumphs of building an AI startup within the competitive tech landscape. Yansen also showcased their product Hippo Scribe. This episode is not just about the technological aspects but also touches on the human elements of entrepreneurship—resilience, adaptation, and the relentless pursuit of solving real problems.

Here are some some episode highlight you won’t want to miss:

  • Journey from Big Tech to Startups: Learn about Yansen’s transition from working at major companies like Facebook and Snapchat to founding his own startup. He discusses the key lessons learned and the hacker culture that influenced his entrepreneurial spirit.

  • The Birth of Hippo Scribe: Discover the origin story of Hippo Scribe, born from the real-world pain points faced by physical therapists in documentation processes. Yansen shares his initial struggles and the pivotal moments that guided his focus towards creating impactful solutions.

  • Leveraging AI for Healthcare: Find out how Hippo Scribe uses cutting-edge AI technology, including tools like ChatGPT and Deepgram, to ease the burden of medical documentation. Yansen explains the unique challenges in healthcare compliance and documentation that his product addresses.

  • The Power of Community and Referrals: A discussion on how community interactions and word-of-mouth referrals have become a cornerstone for acquiring new users without traditional outbound sales strategies.

Fun Fact: During Yansen’s time at Snapchat, he was part of the team that developed the Discover feature, which he noted is now used by more than 200 million people daily and was widely imitated by other social media companies, describing Snapchat as the "innovation lab for Silicon Valley.”

Show Notes:

00:00 Intro to Yansen and his background.
04:31 Startup founder faces challenges and needs resilience.
06:51 Talking less, listening more, improved sales.
11:37 Understanding the challenges of community budget allocation.
13:40 Assistance with paperwork and business start-up challenges.
19:12 Focused on improving product and sales strategies.
21:03 Understanding unique user problems for clinic owners.
24:50 Clinicians work to solve problems for patients.

More Quotes from Yansen:

“The reality of startup is, unless you have something incredible, most of the time people don't really care. So it's really up to the founder to go out, get 100 no's before getting the first yes. That lead to the second yes.”

— Yansen Zhou

“It's two track of effort that we're making. One is, better sales, but two is also keep iterating on the product. We're just so laser focused on solving the problem, which is documentation is really, really difficult for physical therapists. So I would say that's actually some of our primary effort, which is like how do we make that entire thing as easy as possible? And on the sales side, we actually don't do any outbound sales at all. Right now we just, we primarily gain user from referral and also from community.”

— Yansen Zhou

“I was trying to build something with community management because that was also during a time where I noticed this trend that more and more companies are starting to build their user base in discord, in slack, or in like online forums.”

— Yansen Zhou

Transcript:

Demetrios:

Welcome to the AI Minds podcast. This is a podcast where we explore the companies of tomorrow built AI. First, I am your host, Demetrios, and this episode is brought to you by Deepgram, the number one speech to text and text to speech API on the Internet today. Trusted by the world's top conversational AI leaders, startups and enterprises like Spotify, Twilio, NASA, and Citibank. Today we're joined by none other than the founder of Hipposcribe, Yansen. How you doing?

Yansen Zhou:

I'm doing great. Thank you so much for having me on. Demetrios.

Demetrios:

Well, I'm very excited to learn about Hipposcribe, but I want to do a quick overview of your story. And I know you were born and grew up in China. You came over to the US, you got your CS degree in software engineering, and then you started to work for the big companies. Are there key learnings that you feel like you took away from your time at Snap, Airbnb and Facebook?

Yansen Zhou:

Yeah, I think there was a lot of key learnings. I think especially at Facebook, you really taught me what it means to move fast. I think as far as big company goes, Facebook still maintained that hacker culture in a lot of the teams where, you know, everyone every day is just trying to move fast, trying to make impact, trying to build features. I think that really, really was impactful to my own startup journey.

Demetrios:

Excellent. Were there any features that you got to see through and put into production that affected millions of people?

Yansen Zhou:

A little bit more than a million, yeah. At Snapchat, I was part of the team that built the discover product, and I think now it's probably used by, I don't know, more than 200 million people every day.

Demetrios:

Yeah. And copied by basically every other company, social company out there.

Yansen Zhou:

Yeah. Snapchat was the innovation lab for Silicon Valley.

Demetrios:

Yeah, exactly. They innovated and then everyone else said, oh, we like that. We'll take that.

Yansen Zhou:

Exactly.

Demetrios:

So you got to be part of that. That is very cool. Now, I know that you have gone through the YC program and Hippo scribe is almost like birthed out of that time. There's.

Yansen Zhou:

Mm hmm.

Demetrios:

You got the founder bug before that, though, is that correct?

Yansen Zhou:

Yes, absolutely.

Demetrios:

Talk to me about why.

Yansen Zhou:

I want to say it's because first, my parents, they were business owner. They run a clinic all their lives. So I just always wanted to, I guess, become them. And when I moved to Silicon Valley, I noticed, like, a lot of my friends becoming founder, I just noticed all these things happening around me, and it's just really cool. I always admire people that, like, can go out, create something and then create something loved by people and solve a real problem. That just sounds so thrilling to me. And I've always wanted to see, hey, like, will I be able to make a product that solve a real problem in love by users? So is that a give it a try?

Demetrios:

That's excellent inspiration. I can only fathom that it was not easy to leave a job at a company like a Facebook or a snap or an Airbnb because you have everything given to you on a silver platter and to go out on your own and live that scrappy startup life. Were there any reality checks when you made the plunge?

Yansen Zhou:

Oh yeah. Oh yeah. Every day. Definitely. I definitely did not realize how difficult this is before doing it. There are two reality check. One is that I have to admit, like, I was just so lucky to be able to work at this big company where like most of the, like, they build this such incredible product that's already used by so many people. And I was so lucky to work with a group of incredibly talented engineers so that, you know, my little bit of impact is able to leverage up because of this organization.

Yansen Zhou:

And I was just so lucky to be able to do that. But in a startup, when you just started on, you're just by yourself, there's no leverage, there's no platform, there's no resources, and you really just have to go out and try to create that. That was the first big reality check of just how easy you had it at these bigger companies. The second one is probably just how, how much people don't really care about your startup. Are you doing this? I would say, yeah. The reality of startup is, unless you have something incredible, most of the time people don't really care. So it's really up to the founder to go out, get 100 no's before getting the first yes. That lead to the second yes.

Yansen Zhou:

That lead to the third yes. But like that hundred nos is quite difficult.

Demetrios:

Yeah. You all of a sudden go from being a software engineer and just having to worry about your code quality and your ability to ship to now having to deal with humans and get very thick skin and learn how to assimilate those no's and recognize that they're not saying no because of you or anything personal. It's just as you mentioned, I really like that people don't care about your startup unless it really is a problem. It solves a problem that they have.

Yansen Zhou:

Exactly that. You phrase it so nicely. People want to use you because of their reason, because it solves a problem for them, not because of, you know, you have this incredible thing.

Demetrios:

Well, it sounds like you have some stories there. Can you break down a few of the reasons why you recognized, or how you recognize that people do not care about your product?

Yansen Zhou:

Sure. Yeah. Maybe I would phrase it a little bit differently because that sounds so adversarial. Yeah, maybe I'll phrase it as like, you know, I went through my learning of how to discover problem. I would say at the beginning, I probably make. I was making all the first time, found a mistake, which is a dream of a brilliant idea. This would definitely change the world, definitely solve a lot of problem. And then I went out talk to user.

Yansen Zhou:

I got there, I got a chance to talk to them throughout the entire meeting. I would just talk about, hey, how brilliant my product is. Hey, don't you think this product is brilliant? And most of the time, you know, they're nice. It's like, yeah, it's great, it's great. But no sales, no traction until I realized with the help of my founder friends that, hey, and also with the help of YC, that when you talk to a user, you really should, most of the time, shut up. Just go there, ask them about the problem, and shut up as much as you can. That was a big switch.

Demetrios:

But now anyone who's read the mom test probably can relate right now. Yeah, I've even read the mom test, and I recognize sometimes when I'm trying to do discovery about one thing or another, I do it horribly wrong. And because I know the mom test, I know that I'm doing it wrong, but that doesn't actually make me change the actions and me do it right. Did you just do it again and again and again and get those reps in? Is that how you feel, like you were able to discover something and learn how to conduct a good discovery meeting?

Yansen Zhou:

I think a good founder have to learn this at some point. And it took me a long time, probably took other people sooner, I think. Yeah. What really changed for me is just like, getting so many nos and like, conducting all these discovery calls so horribly for such a long period of time, where the pain so severe that it forced me to change. Yeah, that's how it clicked for me. But I hope other people can arrive to this conclusion sooner.

Demetrios:

Yeah, pain sometimes is the greatest teacher. I can relate with that 100%. So you mentioned you kind of stumbled through this founder journey for a while. What was that like? What were you trying to create and what were you trying to pitch people on and them saying no to you or being nice and saying yes, but then ghosting you after the call.

Yansen Zhou:

Yeah, exactly. Try a lot of ideas. Wow, it's been a while, I think.

Demetrios:

And you probably are. You have a little PTSD, so you try not to remember that part. I know. That's what happens with me, maybe.

Yansen Zhou:

Oh, I'd love to learn more about what were you trying to do as well?

Demetrios:

Yeah, I'll give you all that information, but first you answer the question.

Yansen Zhou:

Okay. Okay. Oh, man. I was trying to avoid it. I. Just kidding. Yeah, I tried to build law stuff. I was trying to build something in trucking.

Yansen Zhou:

I had this brilliant idea where, you know, in the us law truckers are pretty lonely when they drive. You know, they drive 10 hours. They in the truck. Oh, wouldn't it be cool to have a social network that was due to the time where clubhouse was all the rage. Like, wouldn't it be nice if there's a clubhouse, you know, there's just a space for trucker where when they're driving, they can hop on and just chat. That was a brilliant idea. Definitely.

Demetrios:

Wait, how did you invalidate that? That is. And I love how you're still holding onto it like it was a brilliant idea. It didn't catch on the way that you thought it would.

Yansen Zhou:

Just don't know that. Honestly, I don't know. To be honest, that idea didn't. I never even launched it. So maybe like, as crazy as sound, maybe if someone made a good effort, maybe it would take off. Yeah, yeah.

Demetrios:

Or some other ones.

Yansen Zhou:

Yeah. I was trying to build something with community management because that was also during a time where I noticed this trend that more and more companies are starting to build their user base in discord, in slack, or in like online forums. Hey, wouldn't it be cool to have a software to manage that? And I still remember this. The dumbest thing I ever did, I built. Oh, sorry. I think this idea was really great and a lot of good company was built on that. But it just reminded me of like, this is like early stumble of a sorta founder. I built a tool that can let slack admin automatically get all the emails of people who join Slack without knowing that that's something that you can already do.

Yansen Zhou:

So I built that and then I email, like hundreds of founders that have a slack community in their company say, hey, wouldn't you want to get an email list of all your users? Anyway, it just bring back memory of the silly, silly attempts.

Demetrios:

Oh, I feel for you. Because that is, as we were talking about earlier, the things that I've done. I have a community of members that are in slack and very much like that is totally something that we can do and we can, then I don't need an extra tool for that. And I think anything, what I've come to think about the community space, especially when it comes to companies and communities, is that companies don't really understand how communities work or what the ROI is on the communities. And so it's very hard to allocate a budget to a tool that's going to help with community. And that community budget, I think if it hadn't, if it didn't open up yet, then I am skeptical it is ever going to open up.

Yansen Zhou:

That makes a lot of sense.

Demetrios:

Yeah. So you discarded the community idea. What were some other ones before you eventually stumbled on hippo scribe or was hipposcribe shortly thereafter?

Yansen Zhou:

Yeah, I would say I actually got into the physical therapy space for a long time before doing hipposcribe. I was, I connected with physical therapist, I connected with a physical therapist during my search for idea and I was just doing the mom test. Hey, what was your top pain point, et cetera? And then I learned a few things about the pt space. One of their top problem, well, they told me two things. Top problem in the US because they have to mostly get reimbursement from the insurance and that money is going down every year. That's the first thing. Two is that the documentation was really heavy. So I was trying to solve the first problem first by building something that can allow them to start their own business quickly.

Yansen Zhou:

Help them take care of some of the paperwork. Involves in starting your own business, which in the US include credentialing, mostly credentialing and then just bringing patient. I try to do that. I didn't think I did a very good job, but that did let me spend like a good few months and talking to just hundreds of users during that time. So even though my attempt for that idea didn't work out, looking back I gained a lot of valuable insight and connections through that time. And then after that didn't work, I was like, okay, just work down the problem list. The second problem is documentation. And I noticed that during this journey.

Yansen Zhou:

Okay, so maybe a little bit more background document. Why that is a problem.

Demetrios:

Yeah.

Yansen Zhou:

So in the US for a lot of healthcare specialties, you, everyone complains about documentation. It's because because of the payers dynamic, a lot of doctors had to see patient back to back all day and then they had to document exactly what they did for every single session. And they have to do it for compliance reason, and they have to do it to get paid at all. And that just really is one of the top reason for burnout. It's one of the top reason why clinicians hate their job.

Demetrios:

And this is after a doctor has spent 8-10 hours looking at patients, then they gotta spend another x amount of hours documenting everything that they've seen. So they have to remember what they saw in the first appointment as well as the last appointment.

Yansen Zhou:

Yeah, yeah. It's very, very hard. And for physical therapists, a lot of the time, this documentation time is unpaid. It's something that you expect to do in your own time at home. And imagine at night, your brain's already fried and you still have to do it. Yeah, that was just really painful. And then also during this journey, I started to notice that a lot of the PT I talked to, they're already trying to solve their own problem with chatgpt. I was like, okay, then maybe AI is a good solution here for this problem.

Demetrios:

And so you saw them using chat GPT to help them fill out the forms.

Yansen Zhou:

They were using chat GPT to do documentation without PHI, without healthcare information, for sure. Yeah, they're using it to generate letters or turning some scribbles into a more coherent documentation, stuff like, yeah, okay.

Demetrios:

And so then hipposcribe was born and you applied to the YC program. Or was that in the YC program that happened?

Yansen Zhou:

Well, this was after the YC program.

Demetrios:

Okay.

Yansen Zhou:

We pivoted a lot during YC and a lot after, until stumbling upon this idea. Oh, fascinating.

Demetrios:

And so then you started, how did you know you hit the right idea?

Yansen Zhou:

I think it's still a lot of time, it still takes a lot of conviction even for this idea. So the first signal is that I talked to more than 200 physical therapists and this is just a coherent mess. This is just a recurring message that we. I hate documentation, I hate documentation. I hate documentation. So it's not a rare problem. And then I started building app to help with that a little bit. I went around, show people, even though I have such strong conviction that this problem exists, building that solution and getting the first user still took two months and do that.

Yansen Zhou:

Two months is like no snows. Nosy. Oh, this is really cool. I want to try it now. Oh, well, yeah, show it to me. No response, etcetera. But at that point, I was so convinced that this problem exists and there is a good solution, I just kept going. And then at some point I started having user to just like I show her the product, she used it and then out of nowhere she just started paying for it.

Yansen Zhou:

I was like, oh, wow. So that's the first 2nd where I was like, hmm, maybe this is something valuable that someone is paying, paying money for it without me asking. Yeah. And then after that we found some traction. So now I feel, I feel more convinced that this is gonna work.

Demetrios:

A little bit of a dopamine hit. I can imagine.

Yansen Zhou:

Yeah, yeah. That 1st. 50, $50 just hit different.

Demetrios:

Yep. I 100% understand. It's like, wow, if you would have told your former self when you were working at Facebook that you could get so excited over $50.

Yansen Zhou:

Mm hmm. Yeah.

Demetrios:

You probably never would have believed it.

Yansen Zhou:

Yeah, yeah. You notice what it means, right? It means like what you build is actually valuable. I think for builders that's the highest form of joy, which is you're validated. Like, your stuff is valuable and that is fantastic.

Demetrios:

So now that you've been on this sales grind for a while, are there some key learnings that you feel like, okay, I understand my process, I understand how to sell it. I can see the messaging better. What have you iterated on to make sure that that traction continues to grow as you go out there and you're talking to people?

Yansen Zhou:

It's two track of effort that we're making. One is, as you said, better sales, but two is also keep iterating on the product. We're just so laser focused on solving the problem, which is documentation is really, really difficult for physical therapists. So I would say that's actually some of our primary effort, which is like how do we make that entire thing as easy as possible? And on the sales side, we actually don't do any outbound sales at all. Yeah. Right now we just, we primarily gain user from referral and also from community. Nice.

Yansen Zhou:

I guess the dots are connected here. Yeah, we have a Facebook community where people ask questions about AI and yeah, we get a lot of interest from that.

Demetrios:

How cool is that? You fight. You were able to connect the dots and make community valuable in the way that you thought.

Yansen Zhou:

Yeah.

Demetrios:

Yes. Nobody, you don't care about the emails, you don't need to scrape those emails, but you still get the value from the community.

Yansen Zhou:

I mean, if only there's a tool that can give me all the emails of all the members that would be so valuable.

Demetrios:

That would be killer. I can bet. So have you noticed that the way that you talk about the problem has changed? Because I hear you talking about how documentation is not fun when you speak to people. Is that mainly what you harp on like, hey, you don't like documentation, we make your life easier. Or are there specific things that you now have a bit more capabilities and ability to articulate the problem?

Yansen Zhou:

I would say this is still the main problem. That documentation sucks. But after talking to so many users, I think you start to gain insight. You started to go deeper and deeper, more details, and you start to understand the unique situation and problem of each individual user. So for us that means that a lot of our users are clinic owners. And that means like, what kind of clinic you are, how many providers do you have and what is the settings like? Do you see patient one on one? Or do you see patient multiple patient at a time? Do you have an individual room or do you have like space where everything's happening all at once? What EMR do you use? EMR is like the software system where they have to enter all the data, where the new doc niche. Which EMR do you use? Because for some EMR we have really good integration to make that process easier. So I think that's the gain here, which is just understanding how things work and try to deeply empathize with each individual user.

Yansen Zhou:

And then, yeah, I don't harp on it on the call, I just listen, I say, hey, you know, tell me about a problem. And as they told me problem. I showed them what we have, asked them if this can solve the problem.

Demetrios:

That's so cool. I'm so excited about what you're doing and to finish it off. I can imagine when the first person that you talked to said, yeah, we use this EMR system. Did you, as your software developer brain, did you think, oh, like EMR or are you sure you don't want to use databricks or.

Yansen Zhou:

Yeah, I definitely feel my engineering bad habit flaring up here and there.

Demetrios:

Well, this has been great, man. I appreciate you coming on here and talking to us. I'm very excited for everything that you're doing and I'm excited that Deepgram can be a part of your journey and help on this. Can you give us a bit of a breakdown before we go on how it actually works and how you are solving these problems?

Yansen Zhou:

And Deepgram has been a crucial piece of the solution. So how it works is the problem is clinicians have to write a lot of documentation at the end of the day. And the solution for that is we have a AI scribe where clinicians can use them either to record the conversation between them and patient or a quick dictation. And Deepgram has been instrumental to turn that audio into high quality transcription. And the quality is so good that our user keep telling us, hey, the program is amazing. I was intentionally trying to mess with it by speaking to the phone next to a music player and they still recognize everything. So thank you Deepgram, for making it so easy to build. Yeah, and after we get the transcription, we, based on their specific documentation template, we use AI to do a first draft.

Yansen Zhou:

And then we have an interface where clinician can review, finalize the notes and then we have a way to enter those data into the EMR for some of them. For the one that we don't support, they have to copy paste. Unfortunately, at the moment.

Demetrios:

Yeah, I imagine that's just a question of time though, because there's only so many EMR systems and you don't strike me as the type that's going to just sit around and wait for them to make an integration into your product.

Yansen Zhou:

Yeah, we, because we just really want to solve this problem. We want to get to the ideal world where clinicians just don't have to worry about this, they can just see patients. That's what clinicians enjoy, that's what this is all about without having to worry. Oh God, now I had to go home and instead of spending time with my family, have sit in front of computer and do this.

Demetrios:

Yeah, yeah. And it is. Data entry is nobody's life's passion or life's calling.

Yansen Zhou:

Yeah.

Demetrios:

I don't think maybe one or two of us out there, but I imagine the majority of clinicians do not wake up in the morning and think, let's go and write some documentation.

Yansen Zhou:

Exactly. They want to treat patients well.

Demetrios:

Thank you for doing this and I wish you the best of luck. I'm very excited for everything that you're doing with hipposcribe. If anybody out there is listening and they're a clinician and they think that hipposcribe can help on their journey to make documentation not suck. Reach out to Yansen.

Yansen Zhou:

Thank you so much, Demetrios. Thank you so much having me on. And thank you, Deepgram, for making such an incredible API with amazing solid support so that we can build this product as fast as we wanted to.