Sequoia

Appier: Creating Impact With AI, One Business at a Time

Artificial Intelligence and machine learning are transforming the world of business intelligence and marketing. But before ‘AI’ became a buzzword, it was not easy to convince businesses that the technology could be a game changer for their enterprise. In 2012, Chih-Han Yu launched Appier with a vision to make AI accessible for all. That vision was tough to sell as it was a technology the world had yet to discover. In the years that followed, Appier pivoted many times on its mission to help companies adopt AI and big data to power business decision-making. On this episode of Moonshot, we speak to Appier co-founder and CEO Chih-Han Yu about what it takes to build a global AI-as-a-service company, and how he tapped on his team’s deep technical roots to build a product focused on customer-centricity.

Watch

Listen

Show Notes

  • Introduction [00:31]
  • Appier’s first-mover advantage [01:26]
  • A race against time to find product-market fit [11:27]
  • Developing software prowess in a hardware kingdom [13:55]
  • Maintain a learning mindset and tap on the best [16:42]
  • Going public: The transition from startup founder to public company CEO [27:00]

Transcript

Introduction

Dewi Fabbri: It was 2010, and Chih-Han Yu was in the final year of his PhD at Harvard. He’d already authored dozens of research papers in Artificial Intelligence, robotics and machine learning. One evening, on the way back from his lab, he experienced a moment that made him realize he wanted to take AI to the masses. Two years later, together with his co-founders Joe Su and Winnie Lee, he launched Appier – an AI-powered platform that helps brands and retailers predict consumer behavior and improve their decision-making.

On this episode of Moonshot, Peak XV Managing Director Abheek Anand chats with Chih-Han about his startup ‘aha!’ moment, how he built a software company in a place famed for hardware, and the challenges he faced building Appier before AI became a buzzword.

This conversation was recorded at Surge, our rapid scale-up program for early-stage startups.

Appier’s first-mover advantage

Abheek Anand: I am thrilled to introduce Chih-Han. But before introducing Chih-Han, I want to share with you guys some context on how Chih-Han and us met. This month, we have entered the 10th year of our partnership. So, we actually partnered with Appier in early 2014… So, this is our 10th year. Chih-Han will tell you guys more about what he’s building, what he has built over the last decade, and more. The company is now a listed company on the Tokyo Stock Exchange. Chih-Han was the original co-founder and CEO. He is still the co-founder and CEO. His other co-founders, Winnie (Wan-Ling “Winnie” Lee) and Joe (Chia-Yung “Joe” Su) are still with him as well. 

So, you know, I think we often talk about the journey of a founder really from idea to IPO and beyond, and one of the things we always talk about is how founders scale in their own personal journey as their business scales as well. And I would say, Chih-Han is probably one of the best examples I can think of, who has really scaled his own personal development as well as he has evolved from being, you know, an incredible technical leader, to now an incredible CEO of a public company. So, we’ll go over all of these things. But I want to share with you a few data points.

I was reading the original memo that we wrote for the Appier investment. Chih-Han, you have not seen that memo. So that’s very interesting actually. So, we partnered with Appier, where we led the Series A in 2014. This was actually one of our first investments in the region. And, we literally met Chih-Han and Winnie and Joe in some casual conversations, as they were… They’d started a business which was… Building a gaming business. Both Chih-Han and Joe were at Harvard; Chih-Han graduated with a PhD from Harvard, and around the time we invested, they had just transitioned from that gaming business into building a company that was, you know, broadly in the marketing and advertising software space. And they were using their knowledge of AI, back in 2013 saying, “Hey, we can actually predict conversions better than most other people can, and this is a better way for brands and marketers to target their audiences, than what they can find elsewhere”. So, one of the interesting things that came out of that conversation for us in hindsight was, we were like, “Hey, this is an incredible AI team. AI is probably going to be big in the future. We should partner with them”. Obviously now is a very interesting time to be saying that because AI is one of the biggest trends that we are seeing. 

The other thing I noticed in that memo was, you know, as in all these memos, we talk about where the company is today and where the company can be in the future. And, we did the math and we said the company was [at] about a million in revenue run rate, maybe a million and a half, back then. And, I was looking at our own models, and our own models, Chih-Han, in the out years, had the company getting to just about $100 million in revenue. How much did you guys do last year? And, can you talk about how much you’ll do this year?

Chih-Han Yu: Oh, yes. I think last year we did… In Japanese yen, I think 19.6 billion Japanese yen, which translates to probably 140 – 150 million US dollars. And this year, we are guiding somewhere around 25 – 26 billion Japanese yen, [which is] roughly 200 million US dollars. 

Abheek Anand: I think this is a massive privilege we’ve had, seeing this business scale. It’s amazing talking to Chih-Han. He’s not just talking about next year; he’s thinking not just about 2023, but also about 2025 and beyond. And so, you know, sometimes when we meet founders in the very early days, it’s very hard to imagine what companies can become over a long period of time. But Chih-Han, I’m going to stop talking and turn it over to you. And maybe a good place to start is just to talk a little bit about the founding journey. And actually, I think there are many interesting things about the founding journey. You started the company with your wife. You started the company when you guys were in the US. It’s a deeply technical founding team. There are so many interesting things about it. Tell us more about the founding journey.  

Chih-Han Yu: Well, thank you. Thank you Abheek. I think as a founder, never get yourself constrained, or allow people to set a boundary around you. And then, yeah, a bit of our story. So, I was actually an AI scientist for almost 10 years between 2000 to 2010. Most of my work was about applying AI in robotics and medical devices. And also, I had the privilege to join a Stanford self-driving cars development program back in 2004, and that car eventually became, what we currently call Waymo, at Google. And, the reason we started the company in around 2010; I still remember, I was about to graduate from my PhD, and frankly speaking, I spent a lot of energy in the last few years of my PhD just to get in a publication; and I’ll be able to look very attractive on the job market to get a professional job. And, I was about to start the interview process for a faculty job, and I was driving to the garage of my apartment, and I parked my car, and I felt, “Wait a second. We built a self-driving car about six or seven years ago. Why am I still driving myself?”

Then, I started pondering on all of the things I had done in the last seven to eight years around AI; and calculated when that technology or innovation can be impactful or can be even used in people’s daily life. And I found that it’s likely to be in 20 to 30 years. Some of the projects I did back then were just too early, and it takes like 20 to 30 years to become mature or make people able to accept concepts, such as a self-driving car. Most people are still driving themselves, [and] ourselves… And so, you can imagine one young researcher who is starting his career, but his project and his work will only be impactful for people after he retires. So, I didn’t want to waste my life, just like that. So, that was the background to start a company.

And, one area that we really wanted to focus on was AI making an impact on the world. And this is very different from the textbook founder story. Usually, you need to solve some problems; you see some clear problems or concrete demands, and then you use technology to solve them. I think that’s a much better way and a much smoother way to start a company. But we just wanted to use AI to drive innovation and impact the world. And then, we started thinking about what kind of idea can really impact businesses now, rather than later, because we didn’t actually start with the demand and problem. So, that actually made us pivot so many times.

One thing we also feel fortunate about is because [we worked on] AI early, there were not too many people involved with AI innovation or any AI-related products. So, we had time to pivot and also learn ourselves [about] what kind of application can be accepted and can be welcomed in the market. And, the first product we did was what we call; currently, there’s the name, ‘AI metaverse engine’. So basically, Abheek says, “It’s a game company”, and I try to say, “It’s not a game company, it’s a Metaverse game company”. So, we built a gaming engine that allows the avatar to learn how the operator interacts with the virtual world, and eventually be able to interact on behalf of a real human being. That was our first product.

This was 15 years earlier, there was big value. And, we tried to sell to many different companies and platforms. And eventually, one game publisher said that, “You guys, apparently, you know data and you know AI, but why don’t you just use your know-how and apply it to consumer data behavior analysis? I have a bunch of user data, and I just want to build a very simple game recommendation widget that can recommend to all my users, the most suitable game for each of them”.

And we saw that it’s just a recommendation engine. It’s really… We felt that it was a really simple task. And, we had built such a complex and sophisticated engine, yet they are asking us to build something so simple. And we spent probably – back in 2012 or 2013two days of coding. It was something we thought was quite straightforward, and then delivered results to the company.

And, I still remember the company still exists and is still our customerand then they came back to us saying, “Wow, your result is actually two times better than what we used to have”. We were very shocked. And then we started asking, “What kind of methodology did you use?” Then we found that the methodology back then being used was really outdated. They just used simple rules and simple statistics. And we found that maybe this is something we really can contribute to the whole ecosystem; tell people to use their consumer data, and then make the best use of it. So, we pivoted our company from a metaverse AI engine to an ad-based consumer behavior prediction platform. So, that was the starting point of Appier.

A race against time to find product-market fit

I’ve been fortunate to start with my two roommates. Actually, one roommate is Joe. Joe was my real roommate that I stayed with back at Harvard University. The other is Winnie. We dated, and then we co-founded the company together. Not like us three founded the company, and then [Winnie and I] started dating each other. We already knew each other. The reason I invited Winnie to join our company is because we were so… Our funding was so constrained. I was trying with the very limited funding that we had; we were trying to recruit someone really smart, who really could help us handle operations. And, I could only think of Winnie. And, I tried to convince her to join our company. But she was very… She had done great in immunology, and published in journals like ‘Nature’. So, she was about to become an immunologist, but she was convinced by me to join the company in year two. So, that’s our founding story.

And, we pivoted many times. We didn’t go out for fundraising. [We] pretty much used our savings to bootstrap the company. We had almost burned through all our savings we had accumulated for the last few years. And so, that was a turning point where we found that our product could be useful for companies. And then, we started by getting some individual projects, and then eventually built a platform and bootstrapped on that. We were literally only two months away from shutting down our company, [before] finally finding product-market fit.

Abheek Anand: Thanks, Chih-Han. I mean, you said a lot of things there. One obviously, you worked on the self-driving car projects before they even were known outside of the non-academic world, the world outside of academia. You worked on a version of the metaverse before “metaverse” was even a word that was used outside.

Chih-Han Yu: Yeah, I only got this word last year.

Abheek Anand: For folks who don’t know, Winnie has a PhD as well in immunology. And so, it’s an incredibly technical team. And then, I remember Joe was winning programming contests all over, and he was a grad student at Harvard as well when you guys were starting off. So, you guys really put together an incredible technical team. How did that work out for you guys? What was it like, building a business in a part of the world, with a team that was deeply technical, but in a part of the world that wasn’t necessarily known for its software prowess?

Developing software prowess in a hardware kingdom

Chih-Han Yu: Yeah, it’s not a very obvious choice; a path people would take. Usually, you’ll find some business people together, and they found a company somewhere where you can find easy access to talent. But back then, because we sort of used first principles… So when we wanted to build a company, the first thing we needed to have was a good technology product. And, because AI was not a hot topic back then, and we didn’t have sufficient funding to really build a technical team in Silicon Valley… But, my co-founder Joe used to be in the national programming team in Taiwan. So, he knew all his teammates, including some of our current founding members, who are still in our company. He felt that we could find great software engineers in Taiwan, and then we came back.

But currently, I think the country is not; or a border is not really a constraint. You can build software somewhere, and re-ship your products anywhere. And we had last-mile service as well. So, back then Apple just launched the App Store. So, a lot of people actually thought that suddenly you could put your app on the App Store, you can distribute and resell it everywhere. Technically it’s not… In reality, it’s not like that. But, if you can build great products, then it will sell itself. We were also kind of naive and thought that it could be possible to be achieved like that.

So, that was the reason we came to Taiwan and built a technical team, and really built the product first. And then, learned everything from product management and basic company operations. In the first few years after the launch of the product, things just skyrocketed and we achieved more than a 100 times our growth rate. So, we didn’t even need to care too much about sales. And at some point in time, scalability became an issue. And, Abheek really gave us the advice that, “You need to really pick up [the skill of] how to run a scalable sales organization”.

So, at each step of our startup journey, we might encounter some constraint or some capability that we currently don’t have. In our case, either we would build it ourselves, learn it ourselves…We used to self-study the subjects. Sometimes we asked for expert opinions. In some areas that were very critical, then we hired someone really great to complement ourselves. But a lot of time if you don’t have the basic knowledge, it is very hard even for you to identify the right talent to join your team. So, having some basic understanding and basic background about each aspect of the company is very important, so that you will be able to identify the right talent to join the company and build a strong team.

Maintain a learning mindset and tap on the best

Abheek Anand: Yes, Chih-Han – and actually, I wanted to talk about sales because I remember we’ve had so many conversations over the years about sales and I remember when you started to first spend your own personal energy on sales, and within a matter of a few quarters, and a few years, I thought you became one of the best sales leaders in the company because you really understood the customers, and you really understood how to then scale sales organizations.

But, that’s not a skill set you had when you started the company. You know, I think a lot of founders over here will also go through that learning journey where they have some strengths, but they need to kind of hire either for people to complement those strengths or learn things that they may not necessarily have as skill sets, to begin with. Talk a little bit about that journey, and I think in particular in sales, because you’ve built an enterprise software company, you’re selling to large enterprises across the region. What was that journey like? I know we had some missteps in the early days, maybe some mis-hires, but over time, sales has become almost a strength for the company. How did you manage to do that?

Chih-Han Yu: Yeah, I think as a founder and especially for a company founded in Taiwan… Our team, we basically had to learn everything from how to run an HR organization, how to run a financial organization, how to run a sales organization, how to market; pretty much every aspect. And, along the way, I think sales are super important for enterprise software companies because even if you build very good software, you still need to distribute it to companies and customers. Especially after priority customers start using your product, you need to be able to have a process to be able to navigate the stakeholder until the end. And, when a customer is stuck while using your product, you also need to have customer success functionality in order to allow them to really feel a good return by using your software.

So, this is a very long process. And I think, in the early days as founders, we had no choice. It was very hard for us to recruit great sales leaders. So, we started by selling ourselves. And along the way, as we were selling, we started getting good revenue, and we got investors, then we stopped bootstrapping, and we hired slightly better sales team members to join us. It is an iterative process; you grow your revenue a little bit, then you are able to afford slightly better talent, and then you go to your next step, and get even better talent in your team. And eventually, build a very good sales organization.

I think for us to really start massive sales is that we had a very good coach who used to run an enterprise software organization with, I think, more than 1000 sales [people]. And, instead of building the process button-up, he taught us how to [do it] top-down and look at the sales organization, and how you should build from [your] end-result, and also back-calculate to the very first step to solve the problem at each step. And to run such an organization, you need to know about how to build a systematic process, and how you can really lock the right numbers. And also use those numbers to help you to scale the organization, and also use those numbers to guide your team and to replicate the capability that you have. I think this is something that we have learnt along the way.

But as founders, especially for a company in a place where there’s almost no software startup company, we really had to learn everything from product management; even how to do international tax. You build your own methodology, and the way to learn things is not that different, because when I did my PhD it was all about studying state of the art [systems], and how people do it. And then, you try to improve it a little bit. We just use that kind of methodology to study how to start a company. But sometimes, you are also constrained by; I always remember that you mentioned, “You don’t know what you don’t know”. Sometimes when you’re in that situation, you don’t know how people run a very scalable sales organization. You have never met such a person, and probably even if you do a lot of research, you are constrained by local optimum. So, that is sometimes a pitfall of doing everything by self-study.

Abheek Anand: Yeah, and actually, we should talk about this topic because I think it’s an important one. I think you touched on this topic of growth and learning through the founder’s journey. And then obviously you guys have seen a lot of that. Let’s talk a little bit about this growth journey. And I remember one of the first conversations I had with Dr Lee-Feng Chien actually, who is our chairman and who was previously one of the leaders of Google in Taiwan. And when he joined us, I remember talking to him saying, “Hey, why did you decide to join Appier?” And I think he joked and said that, “Appier was literally the only company in Taiwan that was able to poach engineers from Google”, which is why he thought, “Maybe they have something interesting going on”.

And so clearly, you know, engineering has been a massive strength for you guys. You’ve had, you know, literally the who’s who of engineering talent in Taiwan, including, you know, professors from top universities join the team. But you guys have also done a really nice job of the things around it, you know, including sales, marketing, finance. And, I think one of the things that you alluded to in your previous answer was this idea of having a learning mindset.

I mean, at the end of the day, a PhD is going through a journey of learning which, you know, candidly, many of those things involve things that maybe nobody knows of before, which is why it’s adopted and it’s actually venturing into new ideas. But you’ve really taken that mindset and applied it to learning about new things, whether it’s transfer pricing, sales organizations, or sales planning, etc. How should founders think about this trade-off between learning things along the way, finding advisors, hiring people for these jobs… What have you seen that has worked for you guys? What has not worked? What are some tips that people can take away, in terms of learning in their own founder’s journey?

Chih-Han Yu: Yeah, that’s a great question. So for us, it’s all about; because we apply the experience that we had during our PhD studies – always learning from the best – so, that’s always a principle. But sometimes, our company in the past was too small to access the best, but we tried our best to access that information. For example, currently, you can ask a lot of questions; you can ask questions to ChatGPT and then you can get some basic knowledge. And based on that you actually chase [the person] who is an expert in that area and find out relevant information based on the concepts, or even have a comfortable conversation with a subject matter expert and then build that capability. 

For us, before starting the company I pretty much only knew AI and mathematical theories. And I pretty much had to build everything on product management in every aspect, myself. But being able to self-study and get basic concepts around one subject is a very important first step, up to the stage that you can then identify who is an expert and then you can go up to talk to the expert. I think this [aspect of] knowing who the subject expert is and always trying to chase the best people to even find five or 10 minutes of [their] time to help you answer the most critical questions – sometimes, it’s just a rational question – can save you a lot of time. So, we always use these kinds of methodologies to do it. 

And, I also spend a lot of time doing interviews. In the early days, in order to prepare to interview talent whose expertise is not in my area, I also studied a lot about the subject that’s relevant to that. But in the interview process, I can also learn a lot. And, by interviewing a few talents, then you probably know 50-60%. Eventually, you will find someone who really knows something you cannot find on the Internet. That’s probably the talent you should hire. And eventually, this person joins the team, and you can learn with this person on a day-to-day basis. I think that’s the best part of being a founder.

Abheek Anand: I think, one of the things that you’ve done very well and I think great founders do very well, is that they never slowed down the rate of learning. It’s easy to do that because, you know, you’re working in a big organization, now you’re running a public company, but the fact that you need to sort of be close to customers, be close to your employees, you know, doing things like interviewing, doing things like making sure you’re always upping the bar on talent, helps everybody stay on their toes, and for your learning journey to continue at the same rate it was when you started.

Chih-Han, I want to spend a little bit of time talking about the next stage; because we talked about the founding journey, we talked about this idea of the growth journey where you’ve sort of learned from people – you hired some incredible people, the business has started to scale. And now, we get to the point maybe a few years ago where you started to consider taking the company public. And, you know, one of the things that you guys have done very well, which I think is super relevant in today’s environment, is always being very capital efficient. I’ve always been blown away by how you’ve always focused on revenue, you’ve always focused on building things frugally, and I think that comes from a culture… It’s a cultural thing. It’s where you guys have not raised a lot of capital, and then you started to consider going public. Talk a little bit about that thought process. What made you think that this was the right time to go and list? What were some of the trade-offs now that you’re a public company CEO for the last several years? What have you learnt running a public company versus a private company? Give us a little bit of a flavor of those trade-offs.

Going public: The transition from startup founder to public company CEO

Chih-Han Yu: I think, roughly at year seven or year eight, up to the point that some people wanted to offer to buy our company because we had certain proven scalability, and also, we started asking ourselves, “Do we want to sell our company, or do we want to continue running the company?” And, we started having this sort of personal discussion among the founders, and it turned out that, assuming that we sell the company and cash out everything, we really don’t know what to do tomorrow. Eventually, we actually drilled down to continue running Appier as the best solution for our lives. And that’s something we really want to do. And, then we decided to go public. And, it was actually not a very straightforward process because Taiwan is a hardware island. So, there’s almost no software company listed here. And, NASDAQ is very far; we didn’t even have a US business. We have business in Japan, Korea and also Southeast Asia. So, we also had to get ourselves listed where most of our people have a domicile. 

And, where is the best place to be listed, was also a very long discussion. For us, we decided, “What kind of life do we want to have?” I think selling the company at the right stage is also a very good choice because that concludes your journey, and then you can start on the next.

In our case, we simply felt that continuing to build this business was the best choice for us. Everyone can have different choices. You build your company to a certain stage and sell the company, that’s great too, because you get yourself rewarded for the product that you have built, and you can join the bigger organizations… And then you can start another innovation. And, being a public company CEO is not easy. Actually, one of the first questions I asked ChatGPT was, “What’s the difference between a private company CEO versus a public company CEO?” It gave a great answer: “Being a public company CEO; you get more scrutiny in the public market, and you have to fulfill the promises and predictions in a much more precise way.

But at the same time, because of this process, your company also becomes much more robust, and your personnel can also learn more”. I thought the answer was so perfect. It exactly matched what we have experienced. Being a public company CEO means that the company is public, it’s not yours, and you really have to be socially responsible and give people visibility about what revenue you can deliver, what profit you can deliver, and also fulfil the promise. And then, at the same time, you also have to make sure everything you do is compliant legally, and also to internal rules.

And also, build a very good organizational structure that can support the continuity of the company, that can sustain for generations. Because, you have to think that the company is public property, not just your own property, or your co-founder’s property, or employees’ property. You have to think that the company needs to sustain itself for a very long term. And, almost think that the company can operate autonomously, even without you. What Sequoia has always mentioned is to build a sustainable business for generations.

Of course, for a young public company, you probably cannot get to that stage within a very short time. But along the way, you have to build that kind of infrastructure to achieve that. I think being a public company CEO has probably been initially challenging, but also a rewarding, learning experience for us in our entrepreneurship journey. But, we are very glad to have many good partners, including my co-founders as well as investors like Peak XV (formerly Sequoia India and SEA) to help us.

Abheek Anand: Chih-Han, it’s been an incredible journey. And I am going to ask you one last question. You talked about the founding journey, you talked about the pivots that you did, when you were in the US, when you guys were just getting the team put together, getting Winnie to join you, and I would say, the journey of going from zero-to-two, you know, product-market fit and a little bit beyond that, then there was that scale-up journey where you talked about hiring the right people, building the right team, learning things that you may not have learnt before – finding advisors, finding board members – getting to a point where the company was scaling at the right pace.

And now you’ve talked a little bit about being a public company, and the challenges and opportunities that you guys are facing at that. You are well into a decade into this journey; and you know, hopefully, there are many more years ahead. But talk a little bit about two things: one, what have you personally enjoyed the most? Which stage did you enjoy the most? And then secondly, what did you wish you’d known over the last several years, that if you had to go back and tell yourself at any point, what would you tell yourself?

Chih-Han Yu: I am still most interested in product innovations. So, whenever there’s a great product design, or product brainstorming session, being someone from a product background, I really enjoy it. And, if it involves some technical brainstorming, that part is the most enjoyable part of my daily life. And I think, also leading a team or identifying great talent who can lead a team, to solve a very challenging task. And then, being able to find a great leader in a certain area that can really help our company scale, I think that’s also the most enjoyable moment for me. Along the way, our founding journey has not been so easy. Initially, for the first few years, AI was not a thing. So, nobody was really interested in what we did but we persisted. Our parents always hinted that, “You probably should get a job. A real job”. 

So, I think there are a lot of moments where you can really be very doubtful about what you are doing. But I’m glad that I don’t need to be in the future to tell myself to persist, because my co-founders, as a team, gave me a very strong belief to persist and also continue on this path. And, we also feel incredibly fortunate that AI took off, starting in 2015, people really focused on it. Now I think people are getting even more interested in AI applications. So I think, studying as an AI scientist or AI engineer, I could not be more fortunate to have this opportunity to really experience this period of time and really build a business to help our customers. And also help our employees to achieve what they could not achieve before.

So, if I have a chance to tell myself, I would just encourage myself to just keep going. But I’m glad that my co-founders back then; we already built a very strong belief to keep going.

Abheek Anand: Amazing. You’re saying, Chih-Han, that AI might not have been a hot area back then… Almost seems hard to imagine right now because there’s so much activity going on in that space, and you guys are in the thick of things. And, I know you’re exploring many new ideas and many new products that you guys are going to launch, so hopefully this itch to be a part of innovative product design will continue to be addressed even as a public company CEO, because you have so many new things that you guys want to be launching. We’ve always talked about it, as a great example of how one builds a high-performing technical organization. I think you guys have done an absolutely incredible job with that. And with that, Chih-Han, I think we should wrap. Thank you so much for doing this. 

Chih-Han Yu: Thank you. Thank you, everyone.

Dewi Fabbri: You’ve been listening to co-founder and CEO of Appier, Chih-Han Yu in conversation with Abheek Anand, Managing Director here at Peak XV. I’m Dewi Fabbri, and for more interesting startup stories, visit our website peakxv.com or follow us on your favorite podcast platform.