The following interview is a conversation we had with Jeong-Suh Choi, CEO Bobidi, on our podcast Category Visionaries. You can view the full episode here: Nearly $6 Million raised to Deliver Better, More Inclusive AI Solutions for Developers.
Jeong-Suh Choi
Choi, thanks for chatting with me today. Hi Brian, thanks for having me today.
Brett
Yeah, no problem. So, before we begin talking about what you’re building, let’s start with a quick summary of who you are and a bit more about your background.
Jeong-Suh Choi
Sure. So, out of 20 years of my career, I can probably have split into two, like first half and second half, ten years each. The first half was more about being in startup world and the second half big court, like three years in Ebay and seven years in Facebook. Now meta. So in the first half I started as a salesperson in a small startup, and then I started running my own businesses in Korea, three of them. And then I sold the last one and I moved to the United States after that for MBA. And then I joined Ebay as a product manager, spent three years there and then joined Facebook as the first product manager at a team called Ad Integrity. And that’s when I learned much about machine learning because the team is all about fighting against the hackers who are trying to exploit the entire ad system at Facebook.
Jeong-Suh Choi
And I learned a lot about how challenging it is to predict and make the machine work in the right way. So I learned a lot about machine learning at that time. And then after that, I joined Social Impact Team for two years in HR technology for one year and back to the jungle in Bobbity.
Brett
Very cool. And take me back. So what year was it that you moved to the United States?
Jeong-Suh Choi
That was 2009.
Brett
Wow, interesting time to move to America.
Jeong-Suh Choi
Exactly. That was a seated there.
Brett
Yeah, that’s fair. So what was going through your mind as you made that shift? Were you terrified of moving to the US? Were you super excited? Walk us through your psychology there.
Jeong-Suh Choi
I mean, both the first two businesses in Korea didn’t go well, but during that time I saw I got to know about Silicon Valley, there’s startup world and all the entire ecosystem. So I was pretty excited about learning about how people work in startup works in Silicon Valley or in the United States in bigger term. So I was pretty excited. But given the economic situation, et cetera, it was terrifying at the same time. But my goal was to stay in the US. And learn, and eventually I can start up a business there. And after ten years, for twelve years to be precise, that happened.
Brett
Amazing. And ebay and Facebook, those are two very big companies, but two very different companies. So just from your perspective in working at both of these large organizations, what was the biggest difference between ebay and facebook?
Jeong-Suh Choi
Wow. I mean, the business model is different, obviously. Yeah, I mean, I learned a lot from those two experiences. In ebay was more about learning the functions of the product manager and then how the ecosystem works. And I was kind of more familiar with ecommerce as an industry back then, so I learned a lot about the skills and hard skills, et cetera. But at Facebook, I started learning more about mission. So everybody in Facebook is obsessed with the mission. They always actually think about the mission and the impact. And I think that was pretty impressive. So I would say, just to summarize, in ebay, I learned about art skills, and in Facebook I learned much about how important it is to think about mission and the perfect execution to make it happen. Nice.
Brett
Two very valuable skills there. And one other thing I wanted to ask you about is your nonprofit, mule forward. Can you tell us about that?
Jeong-Suh Choi
Yeah. So it just started as a casual, personal fundraiser at Facebook right at the beginning of the pandemic, back in 2020, I think it was. In March, I saw a blog post saying that they raised fund and donated the 100% of the fund raised to the local restaurants, which were having a hard time because of shelter in place back then in the Bay Area, so that the restaurants can cook great meals and deliver that to the healthcare workers who are struggling. At the very beginning of pandemic. And they thought it was really great idea, but the challenge they had was it’s cumbersome. It’s so hard to organize all these meals and deliveries and et cetera, somebody has to automate it. And then when I saw the word automation and I was like, yeah, really? In silicon Valley, we can probably do some automation there, build some products.
Jeong-Suh Choi
So I started building the product brief too with the hope that we can automate the entire process. It kind of led to creating a nonprofit meal forward, and we started delivering ourselves with tens of partner restaurants and beneficiaries. Yeah, so far we’ve delivered about like 20,000 still going on in the bay area. And then we are serving now mainly the hungry kids in the bay area. Believe it or not, there are a lot of hungry kids in the bay area, so it’s still going on. Anybody’s interested in volunteering? It’s 100% volunteering based. Please, dave me up.
Brett
Amazing. I see on the website it said like 18,000 meals and almost a quarter million dollars in funds raised so far.
Jeong-Suh Choi
Yeah.
Brett
Wow, that’s amazing.
Jeong-Suh Choi
Congratulations. Thank you. We should do more.
Brett
There’s always more to do, right?
Jeong-Suh Choi
Yeah.
Brett
Nice. Well, that’s helpful background for us to understand where you come from. Really. Now, two questions we like to ask to better understand what makes you tick as a Founder and as an entrepreneur. Is there a specific CEO that you admire the most? And if so, who is it?
Jeong-Suh Choi
And what do you admire about that in general? Not just one, but there’s so many founders that I admire. When you’re in early career, you learn a lot from one person or a.
Brett
Small number of people.
Jeong-Suh Choi
But as you grow in your career, you tend to learn little by little from many people so that you can make your learning curve, in aggregate, really steep. So I’m in that stage, I would say, and I admire all founders, I mean, it’s a really tough job, but if I have to pick one, I would probably pick Brian Chesky from Airbnb. Why? Grit. Everybody knows his cockroach stories at the very beginning of Airbnb. And then he constantly raises the bar to realize something that he imagines, which is beyond what people could imagine. Like, he talks about one day five stars experiences and six star experiences. Seven star experiences to actually make it happen. And last but not least, how he handled the setback. Like the layoff that made at the beginning of the pandemic. But the way that he dealt with the setback was pretty impressive.
Jeong-Suh Choi
There are many reasons, but yeah, I would pick Brian Chesky if I have to pick one. Nice.
Brett
That’s a great call out. And I don’t remember where I heard that, but I’ve heard his exercise of going through what it would look like for a four star, five star, six star. And I think at one point it was like what was ten stars? Like Elon Musk meets you as you fly in on a private jet and drives you directly to your airbnb. Something crazy like that.
Jeong-Suh Choi
Exactly. I think it’s going to happen at some point. If it’s fry Jestic, it’s true. And what about books?
Brett
Is there a specific book that’s had a major impact on you as a Founder? And this can be a business book or could also just be a personal book that’s really influenced how you view the world?
Jeong-Suh Choi
Yeah, I mean, it’s pretty much the same as the Founder that I admire. There are many in many different perspectives, but I can pick Give and Take by Adam Grant. The book is all about paying more attention to what other people need from me and give more to be able to succeed. I’m pretty sure there’s a better version of summary, but it’s summary there. And then when you think about that as a Founder, you just can’t build a great business by just taking. I mean, you need to give meaning. Like, you need to understand what your team needs more in order to build a strong and formidable team and that we all together can make huge impact. So I think that’s really important to understand how to give and make sure the team is actually working as a team. So that was pretty impactful as a Founder to learn about.
Jeong-Suh Choi
And also that is actually in light with what I learned at Facebook or Meta as well. They’re saying that to make a happy customer or user, there are three things people, process and product. And usually you spend a lot of time on product and some time in process and little time in people. But it has to be exactly the opposite. You should spend a lot of time on people, sometimes in process, little on product, and you will see a magical moment of people creating great product to scale. So I learned a lot about that in Facebook, and then that’s kind of in line with what give and take says. So I would actually say give and take was pretty impactful for me. Nice.
Brett
I love that. That’s been on my list for a long time, but you just pushed it to the top. I’ve been meaning to read that for.
Jeong-Suh Choi
Got to really highly recommend it.
Brett
Nice. Amazing.
Jeong-Suh Choi
Cool.
Brett
Let’s switch gears here. And now let’s finally talk about what you’re building today. So can you take us through the origin story behind the company?
Jeong-Suh Choi
Sure. So it was back in 2020, holiday season. I used to write business plan every holiday season, so it was 2020. I thought about this idea and reached out to my friend Sue Jiang, who’s my Co-Founder, but he used to be my drinking friend. So every season, like holiday season, we drink together. I talk about business plan and ideas, and he’s like, okay, let’s just keep drinking. But this time, it was like after five minutes, he said, hey, Troy, this is great. Let’s do this. And I was like, what do you mean by do this? And six months passed. We kind of explored and tested the idea. We started seeing some signals. So back in June, in 2021, after six months after the ideation, we started the company. That’s how we started.
Brett
Amazing. And what are customers paying you to solve? How do they articulate that problem?
Jeong-Suh Choi
So the problem we’re going after is that the AI companies are the customers, and they don’t really know how their AI models will perform in the real world before they deploy. Just to give an example, like Facebook, AI categorizing black people as primates. I feel bad whenever I talk about this. Or Twitter facial cropping algorithm favoring white, younger, female. Or Stanford research that says the automatic speech recognition system they don’t understand, like black accent. There are so many bias that is actually happening, but that’s not driven by exactly the lack of data. Facebook and Twitter, they have a lot of data. It’s actually driven more by the unknown. The unknown bias. So that’s the problem. We’re going after. And the solution that we’re proposing is that, hey, we have our global community where people can find a loophole of your AI immediately. So just to give an example, if there’s an automatic speech recognition model that’s supposed to transcribe whatever you are saying, you can record and let the AI create a transcript and see the transcript and edit all these things that the AI got wrong, like newly coined words, acronyms.
Jeong-Suh Choi
And then you share all this data so that the AI can focus on training based upon what they got wrong so that we can improve the model really efficiently. So that’s the solution that basically we’re providing for. And then now we’re focused on we started as a company covering computer vision, NLP and speech, but now we’re focused on speech data and speech modeling. And we sell like prepackaged test data for speech, mostly in English and Korean. There are really high quality, a lot of metadata, many attributes like text and noise and accent, et cetera. And we also help our customers understand the bias the model has, what kind of data they need to make the model more robust very quickly. Because basically what we do is we connect their models to our community directly and let our communities find the poll for them. So we get a lot of insights very quickly.
Jeong-Suh Choi
That’s what we basically do.
Brett
Interesting. So it’s kind of like a bug bounty program.
Jeong-Suh Choi
Exactly. We call ourselves as a bug bounty platform, as a service for AI. Nice.
Brett
That’s very cool. I love that. And Chat GPT came out like a month ago and was going viral on LinkedIn and everywhere else. Have investors just been blowing you up the last month or so?
Jeong-Suh Choi
Yeah, I would say that 2022 will be remembered as a transitioning year in many perspectives.
Brett
It’s amazing. From your perspective, how would you summarize just the state of artificial intelligence today if we’re looking at the landscape and the ecosystem?
Jeong-Suh Choi
Yeah, I mean, that’s a great question. When you step back, the concept of AI was born back in 1950s and for the first 70 years or so, I think there’s one question that entire industry has been trying to answer, which is, hey, there’s a concept of AI, is it going to work? And thanks to the drastic improvement in data and the data infrastructure and then the sheer amount of data, we can probably see that AI is kind of working. It’s great. And then what’s the next set of or single question that we need to answer in the industry for the next couple of decades? Some VC actually said the question is so what? I mean, okay, AI is working, but what’s the value? What’s the net positive or negative fill? What’s the business impact that the AI is creating itself? Like is it creating more revenue or is it cutting cost or is it helping any kind of leading indicator like customer engagement or user time spans or et cetera, or in the other side, isn’t it damaging the entire community with all this unknown bias, et cetera.
Jeong-Suh Choi
So are we answering this question? I mean, I think for the next couple of decades it is going to be super important for us to answer that question. What’s the net value or net positive or net negative value that the AI is creating and then augmenting the positive value and mitigating the negative value will be a super important part. And the reason that I said at the very beginning, a couple of minutes back, I said that the 2020 or chat GPT, et cetera are going to be recorded as a transitioning breakthrough. I think the reason is because they are creating a business impact like you probably have heard, chat, GPT, Generative, AI, stable diffusion, noble AI, text image, there are a lot of things. But what’s going to happen? Two things are going to happen in my mind. One, there’s a large language model and on top of that there will be a lot of different verticals shooting for different use cases.
Jeong-Suh Choi
Like if it’s a chat bot, there are chat bots to care people or chatbots to hang out with like friends. So it may actually allow some swearings because between friends you swear a lot but you don’t want to actually let your carrying chatbot swear. So there are different use cases. You want to improve your model on top of the large language model. So you will need a lot of data, but the data should be proprietary. It’s got to be legally safe. But it’s extremely hard to ensure that you obtain all this proprietary legally safe data in the industry. Like stable division. A couple of days back they got just sued. So that’s going to be a strong challenge. And the other one is because it’s verticals, got to be able to use or iterate and improve your model very quickly against the competition, not even every month, but also even like every week or even every day.
Jeong-Suh Choi
So how do you secure all this great quality data in a legally safe way and then iterate your model very quickly? These two are going to be a super big challenge for the entire industry. My point of view, and we can help.
Brett
Amazing. Love that. And on your website I see at the bottom there it says start your new ten minute side hustle.
Jeong-Suh Choi
So can you talk us through who.
Brett
Are those people who are using this as a side hustle or what’s the vision for who do you envision is going to be using this as a way to generate income as a side hustle?
Jeong-Suh Choi
Great question. Our basic premise is that the best people to collect the data for you and validate if your model is actually working or not is actually the people you are targeting with the service that’s backed by the model you’re building. So that’s the basic premise. So it’s global, public, general people mostly in the US. Korea, Philippines, India, Kenya, South Africa, et cetera. That’s what we have right now. And then that’s in line with the vision that we have. We want the AI to be run by the entire global community with no bias. And the best people to actually say or tell us that the AI has some opportunity to improve in terms of the bias will be the people in the global community. So that’s the vision that we have.
Brett
And what do you envision as a rate that someone could earn if they were doing this? Have you thought through that or worked through what that kind of business model would look like?
Jeong-Suh Choi
Sure. People can earn about ten to $20 an hour easy at Bobidio right now. And when you look at the minimum wages in different communities, like in the Philippines, for example, the minimum wage is about one dollars. And like in Kenya, it’s just $0.28. So there’s a huge arbitrage. So people just love using Bobidi and earning side money as a side hustle. But that’s just the beginning. We understand that the reward and all the side hustle strategy, it’s not going to be enough for us to cover the entire spectrum of the entire community that we have in the Earth. So while we’re starting with utility, just giving rewards at the very beginning, but we have a strategy to grow our community as more social capital based later on. Like, you have identity, have fun, it’s gamified, you have your history here, you have your expertise, et cetera.
Jeong-Suh Choi
I think that’s the way that we can grow the community in the right way.
Brett
Amazing. I love it. And as I’m sure you’ve seen over the years, and especially in the last few months, there is just a lot of noise around AI and a lot of hype around AI. What are you doing to really stand out and separate yourself from all of that noise and all of that buzz? This can be kind of two buckets. It could be, how did you stand out and break through the noise with investors? And then how are you standing out and breaking through the noise with customers?
Jeong-Suh Choi
Wow. That’s a huge question for the investors. I talk a lot about the market shift because we always say that the big data era is ending. Because if there’s a problem with the AI, industry tends to put even more data, big data, so to speak, but it’s not only inefficient and expensive, but also it may augment the problem the data set already had. So you just don’t know. So our approach or our unique perspective around the market is shifting big data areas and the quality data area is coming, and that’s where we’re playing. And the unique approach that we have is also helping us stand out among all this noise. Like, for example, we have a unique approach for a huge problem. We’re not just collecting data, we connect our customers data to our community so that we can find a loophole and that we create the data universe that the customer didn’t know before.
Jeong-Suh Choi
So the approach is unorthodox. And then I think that the investors are pretty interested in all these unorthodox approaches. And last but not least, we talked a lot about the Strong team. We started as two co-founders. But as you can see from the idea, it’s a mashup between community and AI. And that represents the expertise that co-founders we have as well. Like, I spent a lot of time in the community. My friend and Co-Founder Sijan is the AI guy. He’s the PhD in AI and Google Apps team and he rental his AR businesses and sold it to Nayanti. So I know those all stories actually help us stand out a little bit during this tough time for the customers. At the end of the day, it’s the quality that we provide, not only high quality that we collect from the users or the communities directly, but also it processes in a way that you don’t have to actually waste anything.
Jeong-Suh Choi
When you usually buy prepackaged data or from your service log or other sources or even crowdsourcing, they ask people to just to follow this instruction and collect the data. And you just don’t know how much or how helpful it is for your model. But we start with not the instruction, but with their model. So we only focus on the opportunities that the model has and we turn that into a data, high quality data and insight. So that’s why customers actually love working with us.
Brett
And as I’m sure you’ve also experienced, bringing innovative technology to market isn’t easy. What would you say has been your greatest challenge so far and how do you overcome that challenge?
Jeong-Suh Choi
I mean, a lot of things are challenging in that perspective. Yeah, I think it’s a great question. I think the innovation is the function of what you do and the outcome. I mean, you can do dramatically less to get to the same outcome, or you do the same and you get just tremendously different outcome. That’s I think really rough definition of innovation. And the mistake that I had at the very beginning of Bobidi was that I focused too much on what we do. It’s cool and the approach is awesome. We started actually tapping out of the market. It kind of attracted the tension from the VCs and the customers. It wasn’t really, I would say, hurried, meaning it didn’t connect with the direct revenue, didn’t connect with the actual outcome. At the very beginning. We talked a lot about edge cases, how the bias could be actually fixed and et cetera.
Jeong-Suh Choi
But the day to day challenge that the people had at the customer level, they understand that’s a challenge. But day to day challenge is a little different. It’s a data collection, the high quality data. And they wanted to understand why their model is actually not working and they wanted to partner with somebody. So, two needs, they needed high quality data and they needed partners to just deep dive and partner with them to improve their model very quickly. So we changed our strategy of the go to market strategy and then started working with the customers for their own day to day needs. So, just to give an example, like one customer working on ASR model, we started collecting the data for them using our community, using our app iOS android. So basically how it worked was they wanted to collect all this English data from people at casual conversations, but not too casual, like talking about just weather.
Jeong-Suh Choi
So we came up with this app that asks our community all these prompts like, what’s your guilty pleasure? Or what is the one thing that you want to change about yourself? Et cetera. They started recording all of these and the recording will be the app is directly connected to the customer’s AI through API. As soon as people record, the AI will transcribe and we ask the people to edit the transcript wherever the AI got it wrong. So we collected all this data and then gave it to our customers. But because we used their AI, we also had access to all these great insights. Like we found some bias. This AI is not recognizing certain accent, or it’s not working with the babbling noise background, or it’s not really working in a conversational setup. So we came up with a new next plan. Like, hey, we can run this challenge to cover the accent, so we can run this in New Zealand, India, and the Philippines.
Jeong-Suh Choi
And then we ask people to record it in cafe with other people so that you can have more babbling noise in a conversational format, so that we can just laser focus on the data set, creating the data set that the customer actually needs. And that’s how we actually working so far.
Brett
That’s amazing. Now, last couple of questions here for you because we’re almost up on time. What excites you most about the work you get to do every day?
Jeong-Suh Choi
Wow, that’s a great question. AI has a high potential. Everybody knows this and as well as a huge risk, so everybody knows the potential, so we can talk about the risk a little more. I mean, in the real world, there’s some level of randomness and diversity. So it’s actually not too bad at this point. Just to give an example, like call center, if I call a call center and the person over the phone doesn’t recognize my accent, like the accent that I have right now, I can just hang up and call again until I see some person who can actually understand me. But imagine that people in the call center are replaced, completely replaced by the AI and AI does not recognize my accent. I’m completely excluded from that ecosystem. You can imagine this happens with like for example, recruiting AI or interview AI. Because of the false positive or false negative or the bias that the AI unconsciously has, a person can actually lose the job forever.
Jeong-Suh Choi
So there’s a huge implication. The stakes are pretty high. So we’re pretty excited about working on it. Not the problem itself, but the fact that we’re working on this really important problem, but in a way that actually helps the global community. We talked about the rewards and the utility and et cetera, but we get a lot of direct messages from people. Like some person just sent me an email that her mom passed, but Bobbini was really mentally and financially helpful, et cetera, where this guy from the Philippines actually messaged directly us that his wife is expecting a baby. But they’re living separately right now, so he’s working really hard and Bobidi right after his work so that he can support the baby later. So he’s grateful for Bobidi. And that’s why he is going to name his baby after Bobidi. Et cetera. Bobidi is a godfather.
Jeong-Suh Choi
It’s a joke. But still, whenever I get these messages I feel like this is amazing. So when I got this message, I told my Co-Founder Sushan that hey, I know the business is going to be hard. Running a startup is challenging, but whatever happens, I think I can do this at least ten years with no problem. I just love this.
Brett
That’s amazing and certainly sounds rewarding to get those types of messages as well.
Jeong-Suh Choi
And if we zoom out into the.
Brett
Future, what’s the five year vision for the company?
Jeong-Suh Choi
The five year vision for the company is to help everybody believe to run their own challenge. So just like any professors writing business or students writing business or data scientists, CTOs or modelers engineers, they can set up their own budget, their own time period, their own target, like twenty S in their college using iPhone, et cetera, and collect the data and validate their model on themselves. Like by themselves running challenges as if like flower shop owners running Facebook ads. I mean I started my career at Facebook in the Ads team as I said. And back then we had this problem on a cold as lightweight interface that’s helping the long tail the general public create the ad. Easy the revenue was back then was about like 15% coming from the long tail. I think seven years fast forward, it grew up to like 80% or so when it quit.
Jeong-Suh Choi
Long tail can be super powerful. I’ve seen it. I’ve built a product and when I see our product to make it available for anybody, I think our product is much simpler than the ad product that the Facebook had to run in many perspectives. So I can imagine that whenever you’re trying to train your. Model or test the model or deploy the model. Everyone should say that, hey, you should do Bobidi before you do whatever you do. Otherwise you’re screwed. If that happens, I think that the business can be really big. So I would love to actually have everybody use Bobidi to make sure that they get the right data and train the data, train the model in the right way. And that’s the vision that we have.
Brett
Wow. I love that. And I also have to say I love your guys name. It’s such a fun name to say.
Jeong-Suh Choi
Thank you.
Brett
No problem. And unfortunately, we are up on time, so we’re going to have to wrap here before we do. If people want to follow along with your journey as you build this company up, where’s the best place for them to go?
Jeong-Suh Choi
Go to Bobbity.com or send me an email to helloed, bobidi.com.
Brett
Amazing. Well, thank you again for taking the time to come and chat here and share your vision for what you’re building. This is super exciting and we look forward to seeing you execute on this vision.
Jeong-Suh Choi
Thank you so much for your time, Brett.
Brett
No problem. Let’s keep in touch. Our channel.