We need to get the health care revolution right.
Artificial Intelligence promises to reduce bottlenecks, improve quality of care, and allow our over-stretched healthcare systems to scale to meet the needs of the aging global population.
But it’s not going to be easy.
Today we talk with Kota Kubo, founder of Ubie about the opportunities and challenges involved in the coming wave of healthcare innovation. And since Ubie just raised $27 million to fund their global expansion, you’ll be hearing a lot more about them in the future.
It’s a great conversation, and I think you’ll enjoy it.
- The right way to use and AI symptom checker
- Is founding a 50/50 startup with an old friend a good idea?
- How you can manage 150 employees without managers
- Why the team designed Ubie’s UI in Hibiya Park
- The “karaoke interface” for medical data
- Why you should ignore your customers and listen to your users about design
- Why it’s so hard to sell to doctors (and how to do it right)
- How to better support orphaned diseases
- Ubie’s strategy for going global.
Links from the Founder
- Everything you ever wanted to know about Ubie
- Check your symptoms with Ubie’s AI Symptom Checker
- AN overview of Teal Management
- Follow Kota on Twitter @quvo_ubie
- Connect on LinkedIn
Welcome to Disrupting Japan, straight talk from Japan’s most successful entrepreneurs.
I’m Tim Romero and thanks for joining me.
Even if AI can’t give us the right answer, sometimes it can help us to ask the right questions. It turns out, that’s a lot more important than you might think.
Today, we sit down with Kota Kubo, the co-founder of Ubie, an AI based symptom checker and hospital check-in tool that is being used at over a thousand hospitals and clinics across Japan. And as you’ll see in this case, the questions, the AI raises are more important than those it answers. And since Ubie just raised 27 million to fund their global expansion, you’ll be hearing a lot more about them soon.
We also talk about how Ubie manages 150 staff with no managers, why it’s so hard to sell to doctors and how to do it right. How to bring attention to orphan diseases and why you really need to ignore your customer’s ideas about UI and listen to your users.
But, you know, Kota tells that story much better than I can. So, let’s get right to the interview.
Tim: So, we’re sitting here with Kota Kubo of Ubie, who is disrupting digital health here in Japan. So, thanks so much for sitting down with me. Health tech is so important in Japan. And so you’ve got two different products you offer.
Tim: So, let’s do just a real quick introduction to what those products are and then we’ll dive deep.
Kota: Yes, we have the two side of product. First, is for the patient product. It’s our AI symptom checker Ubie and the users input their symptoms like headache or stomach ache or something. So AI asked some of the sort of questions. And after that AI suggest a disease name so their users can get to know their symptom, condition and disease. And after that, we also suggested the clinics or hospitals.
Tim: Okay. You know, one thing I’m curious about that, because I’ve used it. It’s really interesting, but so like some sites like WebMD, for example, they have a really famous problem where someone will go on with like, I don’t know, a runny nose and they’ll start searching and asking questions. And 10 minutes later they’re convinced they have like rabies or some brain eating parasite or some horrible disease. How do you stop that kind of unhealthy interaction at Ubie?
Kota: Yeah, it’s very difficult. I think so our end the point of the suggestions guide people to the appropriate clinics and hospitals and we suggest a specialist created to their symptom. So, I think their users feel they’re safe.
Tim: So, and again, you’re really focusing on providing information, not diagnosing anything.
Kota: Yes. Yes. It’s a very important point.
Tim: And your other product is the monshin, the check in functionality, right?
Kota: Yes, exactly. So, second product is Ubie, health assistant it’s called AI Monshin in Japan. The objective is to reduce the burden of their medical doctors. We save the time of their making the clinical record because our AI also ask a lot of questions about symptoms and AI summarize their medical record.
Tim: This is really cool. So, let’s take time to walk through the process. So, if someone’s got an appointment at the hospital or at a clinic, they can check in and fill out the symptom forms at home on a mobile phone, right?
Tim: And where does the AI come in?
Kota: There a lot of questions from the AI and the AI is changed the question according to their patient answer. So, AI selected the question and also calculates their disease probabilities.
Tim: So, what the AI’s doing is it’s figuring out the kinds of follow up questions a doctor would ask.
Tim: And getting this information together and presenting it to the doctor all at once?
Kota: Yes. Yes.
Tim: So again, not diagnosing, but just getting the information together.
Kota: Yes. Yes.
Tim: Awesome. So, tell me about your customers. Who’s using Ubie and monshin?
Kota: First, is a medical doctors. Before using the Ubie, they have to make their medical record by themselves from scratch, but we support them to make the record. Before we are face to face medical consultation with their patient, also we suggest disease to the medical doctors. It prevents their doctors from overlooking the serious disease.
Tim: So Ubies is use in all 47 prefectures and 15% of all clinics and hospitals in Japan are using it now.
Kota: Yes. Around 15%. That’s the product that they use is either not the Ubie AI Monshin, it’s a Ubie link. It’s a connection between the B2C and the B2B app and in terms of the Ubie AI health assistant, we have just install the 1000 clinics and hospitals.
Tim: That’s fantastic. Okay, I want to dive deep into the product in the market, but before we do that, let’s back up a little bit and talk about sort of how you got here. So, you officially founded Ubie back in 2017, right?
Tim: And at that time, you were still a student?
Kota: No. Before starting Ubie, I worked through Japanese data health company. I worked as a marketing engineer. Basically I did a light code, production code and also I do the marketing related things.
Tim: But that was relatively briefly, right?
Kota: Yes. And including the internship just three years.
Tim: And your co-founder Abe-san, he was also at University of Tokyo, but you guys were like high school classmates or friends or something, right?
Tim: How did that happen? Did you reconnect in university or did you stay friends the whole time?
Kota: Yes. He was my friend and from the high school, as you mentioned, we are friend of there just starting together. So, for the Jyuken. And we split when I entered the university and my bachelor’s degree was in Kyoto University, but he entered the University of Tokyo. And from there graduate school, I moved to the university, the Tokyo and yeah…
Tim: And reconnected.
Kota: Yeah. We connected. And actually when I was in Kyoto University, there was angel investors who liked Abe-san company and his mission is to increase the number of students who want to become an entrepreneur.
Tim: That’s my mission too.
Kota: Sounds nice. So yeah, I got interested in such entrepreneurship and also they’re starting business.
Tim: So Abe-san, I mean, he’s a medical doctor.
Tim: Did you pitch the idea to him or did you come up with it together?
Kota: Yes. In 2012, I come up with the idea of diagnosis. I researched about such domains and I also developed a better version. I kept one year. And after that I understood, I lacked the knowledge of such a medical domain so I have to consult with him. So, I pitched.
Tim: You pitched.
Kota: I pitched him and he said, we can make it.
Tim: So, you were actually working on this from 2012?
Tim: So, you were developing it for like five years before you started Ubie?
Kota: Yes. Yes. That’s a beginning phase is just for the research.
Tim: It’s interesting. So you and Abe-san are co-CEOs?
Tim: Now this is something that’s actually pretty common in Japan but pretty rare in the US. Does that ever cause problems? How does that work on the personal day to day? We’ve got to make these decisions under high stress situation.
Kota: Yes. it’s a very good questions. Yeah, sometimes I feel it’s confusing for the employees too. So, who is the final decision maker? They don’t know. Yeah, I think there good point is we can advise the risk and there is a rules are between me and my co-founder. If there is an inevitable conflict between us after discussion, we can make the same decision.
Tim: I like that. I mean, it’s a very culturally different approach having 50-50 forces you to reach an agreement. I’ve always said in the deals I’ve done, like I’m perfectly happy with 49% or 51%, but I never want 50 because I don’t like the ambiguity, but it is interesting because it does force you to sit down and get aligned.
Kota: Yeah, yeah, yes. Yeah. Some of these investors advised us don’t go for 50 and 50 too.
Tim: Give someone 51 and…
Kota: Yeah, yeah. It’s our entrepreneur.
Tim: I mean it’s different. I mean, what I would imagine would happen is it would definitely slow down decision making, but maybe those are the cases where it’s good to slow down a bit and build a consensus.
Kota: Yes. And we have also installed the framework of the sulture structurings it’s a holocracy, it’s a kind of Teal organization. You know Teal organization?
Tim: I do. And I had a note to talk about this later, but let’s talk about it now. I think the way you’ve structured your organization is super interesting. It is flat to the extreme because it really is unusual and unique and very Ubie.
Kota: Yes. More confusing. But our companies are private organizations, one of their organization adopted holocracy framework, the Teal organization. And in that framework role are the very clear, role and responsibilities. It’s a framework for the delegation from the CEOs. So, we can delegate some of the roles of CEO to each members.
Tim: And to be clear, I mean, Ubie is no longer a small company.
Tim: In fact, an interview, I was reading of you guys last week. It was saying at Ubie we have no managers in an organization of 200 members. And I made a note under that saying what???
Kota: Yeah, it’s very there.
Tim: How does that work in practice? I mean, that’s pretty flat.
Kota: To be honest with three organizations. As I mentioned, we have the five and two organizations have such a hierarchical structure. So, there is a manager actually who is a true holocracy organization. So, there is no manager, but there is a hierarchy, but it’s just a relationship between the low hierarchical there. There is a lead it’s like a manager but these don’t have their accountability of the people management.
Tim: Oh, I see. Okay. But for example, among the developers, are there traditional this is a six person team with one lead managing the team or is it flatter than that?
Kota: Yeah, there always a lot of the product manager or product owner and product owners, accountability includes prioritization of the background. However, there is no titles and positions.
Tim: Oh, okay. I see. So, you’ve separated the functional roles from the management hierarchy?
Kota: Yes. Yes.
Tim: So, some people have responsibilities for the team and they make decisions for the team. But they don’t necessarily supervise or control that team?
Kota: Yes. Yes.
Tim: That’s very interesting. Well, definitely put some links up on the website about that because I know a lot of people are going to want more information about it. But what you’re doing is definitely working. You just raised a little over 27 million, congratulations.
Kota: Thank you very much.
Tim: As always, when you raise that much money now comes a hard part. Let’s dive a little bit deeper into the product of the business. So, you always describe Ubie as physician supervised symptom checking. So, what does that mean in practice, physician supervised?
Kota: Our product is AI and rule based. And we use probability between symptom and disease. Like if a person have their influenza, the probability of having the fever is more than 95%. We use such a probability value for that. And also AI ask some of those question. And after that suggested disease to the each users and including the medical doctors and the medical doctor checkup suggestion is a correct or sometimes not correct.
Tim: Okay. And that provides additional feedback and training.
Kota: Yes. Yes. Awesome. One of through accountability of the doctors to generate such a data.
Tim: That makes sense. And also at one point you had to redesign your entire UI to make it more friendly for elderly users.
Kota: Yes. Yeah.
Tim: Tell me about that process because it was really interesting.
Kota: Yes. So, well our first we provide their tablet to interface for elder people. Sometimes they don’t have their own smartphone. So, at that time patient has to input their symptoms. At first we use their iPads default keyboard, but it’s a very difficult for the patient or elderly people to input and we converted it to our own keyboard card like ATM, ATM is a bank account ATM.
Tim: But you actually went out, you took these tablets out Hibiya Park.
Kota: Yes. Yes. Just test whether it can be available for people to use.
Tim: So, you’re just walking up to elderly people who are sitting on benches and saying this, can you input your symptoms here?
Tim: How did they react?
Kota: Reaction is we could see it’s available from the elderly people. Yeah. First feedback before testing that we heard from the medical staff, it’s impossible to use for the elder people. Actually their first version of our product, the simple select place, we display the human body and they have to drill down. But we’ve felt it’s not like a Google style. It’s like a Yahoo style, previous Yahoo directory style, not the search style, but we thought it’s better to search even for the elderly people. Its better you ask for them to search with a keyword. So, we test it in park.
Tim: And was that the case? They preferred the keywords?
Kota: Yes. Yeah. We search about such color case interface. It’s like a ATM you right. The people use such a color case.
Tim: Oh right. So, modeling it after the old karaoke machines.
Kota: Yes. Yes. Karaoke machines.
Tim: No, that’s beautiful. You use what your customers know.
Kota: Yeah. Yeah. They used to that.
Tim: No, I love that because if you could have interviewed doctors and nurses and hospital administrators and they would never tell you that they would tell you something completely different than going out and talking to people in the park.
Kota: Yes. Yes.
Tim: So, what is your billing model? I mean, it’s a SaaS product, but do you bill by the number of appointments, do you bill by the size of the clinic?
Kota: Our patient model SaaS is related to the number of their medical rooms.
Tim: So, it’s a fixed monthly cost based on basically the size and capacity of the clinic.
Kota: Yes. Yes.
Tim: And what’s your go to market? How do you acquire customers? Selling to doctors is notoriously difficult.
Kota: Yes, exactly. Yeah, it was very hard actually before Ubie penetration, it’s very difficult to introduce such a cloud based service into the hospitals. It was very hard because their healthcare systems are on premise services. So, they cloud based service. The understood as they’re very dangerous, not secure.
Tim: Is that just the perception or there some like regulations that require…?
Kota: There is a guideline of the health care information system on parameter recommended, but we have to keep our line with such a guideline. So, we partnered with the NTT, the meds closed our circular connection, like a BPN. Yeah. After that we can penetrate. Actually our marketing channel in the beginning of that worked well, was just writing a real letter by my co-founders peace medical doctors and send it to the managers of big hospitals.
Tim: Sending it by post?
Kota: Yes, yes. But other people did the operation. It worked well. The good point of our marketing was also our government procedure innovation says how to work. We can get such a good trend. They want to renovate how to work because medical doctors work too much.
Tim: So, it sounds like the key selling point was that Abe-san the medical doctor was reaching out to other MDs saying this is trustworthy. It’ll save you time.
Kota: Yes. Yes.
Tim: I think you have to do that. The only way to get through to doctors is to have other doctors, they just won’t listen to anyone else. Actually one of the things I’ve heard you and Abe-san, talk about is one of your objectives is to better support and treat orphan diseases.
Kota: It’s led to another side of the business. Some of the clients of the pharmaceutical companies in terms of the orphan disease, the disease are very difficult to diagnose for medical doctors who are not specialized for such disease. So, problem for pharmaceutical companies is if there is a patient of the overrun disease and the lay disease, sometimes doctors overlook that. We input a lot of type of disease information so we can suggest the possibility of their orphan disease to their patient and also provides information of their visual race clinic.
Tim: So, ideally even if it’s an orphan disease, there might be a lot of people who are undiagnosed. And so Ubie can help draw attention to the possibility that they might have this disease. It could raise awareness among the doctors that, oh, this person it’s worth looking into and then the market might be big enough for the pharmaceutical companies to invest in treatment or cures. All right.
Tim: Is that happening? It sounds awesome in theory, are we moving in that direction? Do we see this happening or is it too early to tell?
Kota: Yes. In some of the project, we can guide people with orphan disease and we have that questionnaire after we suggest that disease and users answer to that suggestion is very correct. We can guide the people to their appropriate medication with such as suggestion. We also suggest the disease to their doctors and sometimes they’re not specialized the overall disease.
Tim: That makes a lot of sense. And you’ve mentioned you’re working more and more with pharma companies recently. Is the pharma mostly interested in these type of orphan diseases and better understanding the size of the market or do they have other interests in working with Ubie?
Kota: Major practice is all foreign disease and specialty disease are not a primary care disease. Also in terms of clinical trials, it’s a difficult for the pharmaceutical carrier companies to gather the patient with the rare disease. So, we can efficiently find such a patient and suggest their patient to having such clinical trials.
Tim: Okay. So now, especially with the new investment, Ubie is going global what are your primary markets outside of Japan?
Kota: Our first priorities are US.
Tim: I know you guys have been sort of planning and overseas expansion since day one. And that’s great. I wish more Japanese founders had that ambition from day one, from day zero, but healthcare is such a heavily regulated industry everywhere. So, how does that affect your market entry strategies?
Kota: Actually, it’s very hard. We have to research a lot about their regulations, but actually in terms of their healthcare industry, their US has their own regulations.
Tim: The HIPAA, right?
Kota: Yeah. HIPAA. It’s very select, but fortunately there are already symptom checkers, so it’s defined, but suppose Japanese market, we are the pioneer. So, we have to make the rule from scratch here really. And we have to have their lobby with their government but in the US, there is a rule, so we have to align.
Tim: So, I guess it makes it easier because at least the rules are defined.
Tim: But when you go to, let’s say the next market would be the EU or would you be sort of reinventing that from scratch again? Or is it the same service you just have to comply with certain data storage requirements? How big of a difference is there from market to market?
Kota: The biggest point of the business model is there pharmaceutical marketing. So, in the US pharmaceutical companies can advertise their drug. It’s a very unique point in Japan. It’s illegal to advertise their prescribed drug.
Tim: That’s a good point. Yeah. The US market, I think, I don’t know, if’s the only one in the world, but it is unusual that that pharma companies can advertise directly to consumers.
Kota: Yeah. There are only two. Two markets that are US and the New Zealand. I’m not sure why New Zealand.
Tim: So, that’s interesting. So, the biggest differences are not regulations or data storages. It’s just market structure and competition and value proposition?
Kota: Yes, yes. And our visions model depend on the pharmaceutical companies, but our competitors do the monetization with insurance companies. So, the visions model different and our competitive point is just suggest general accuracy. So, we invest a lot of money to develop high accuracy disease suggestions, and we can also acquire training data in their clinical field. It’s also a unique point.
Tim: You know, getting back to that question of regulation in Japan, in particular, well, everywhere really, but in Japan, like healthcare is such a heavily regulated market. In a lot of ways there’s only one buyer, right? The national health insurance system.
Kota: Yes. Yes.
Tim: In what ways does that make innovation hard? In what ways does that make innovation easier?
Kota: I think there are good point and also difficult point.
Tim: So, what are some of the good points?
Kota: Good point is it’s a very simple in the US or other markets, insurance companies pay for cost of their medication, but in Japan, government pay for that. Actually the government moving is a very slow, so it’s very difficult for the startups to seeing the monetization point to government. So, some of the startup main devices observe like AI or software devices, but there another ways pharmaceutical companies, because the Japanese market has a big pharmaceutical market needs. Monetization point is pharmaceutical companies, sales and marketing. It’s one of the unique point of the Japanese market. So, the pharmaceutical companies become a good monetization point.
Tim: That makes sense. They’re constrained by regulation. They have the money to invest, but yeah, it’s just health tech in Japan for the future of Japan, it’s just so important. I mean, everything from telemedicine, I mean, obviously the pharmaceuticals are a big part of it. Technology like Ubie, which allows individual doctors to make a bigger impact. It’s so important. What should Japan be doing to encourage more health tech innovation?
Kota: Actually Japanese, such a traditional power is a very complicated in terms of the telemedicine. So, very slowly penetrate in Japan. I think very although industry problems over years, but there are uniqueness of their Japanese market is their pharmaceutical company pay a lot cost of medication, very high and increasing. And there are a lot of problem, like wholesale of their drugs. And also the government efforts are flexible doing the Ubie business. So, they are very negotiable around the researching their overseas.
Tim: Yeah. They’ve recently set up the sandbox.
Kota: Yeah, yeah.
Tim: Sandbox to allow startups to experiment a little bit, but like safely under close supervision. So, I mean, things seem to be headed in the right direction.
Kota: Before diving into this industry, it seems to be very hard to enter into the market, but industry people also have a lot of concern and they want to solve the problem and that they now thinking they’re such a growing industry.
Tim: Well, and that’s the amazing thing. Like literally everyone I talk to from VCs to founders, to government bureaucrats, to politicians, everyone wants more med tech startups.
Kota: Yeah. Yeah. But not only for the medical doctors can start this industry’s business. I think we can learn. Yeah, of course we need the support of their medical professionals. That’s it’s easier than expected.
Tim: I see really simpler than expected. That’s good to hear.
Kota: Yeah. Just removes such a image of such difficulties.
Tim: That’s refreshing to hear if the biggest problem is the image that it just sounds scary, that’s a solvable problem. Hey, listen, Kota, before I let you go. I want to ask you what I call my magic wand question and that is if I gave you a magic wand and I told you that you could change one thing about Japan, anything at all, the education system, the way people think about risk, regulations. I mean anything at all to make it better for startups in Japan, what would you change?
Kota: Interesting. I want to change the younger people’s language to the English.
Tim: Why is that? Why would you want to change that so everyone’s speaking English?
Kota: We lack the example of the successful startups. I love to laser the book overseas startups as successful cases.
Tim: Would being able to speak and read English, it would give young people more access to information about startups and let them hear more stories, success stories?
Kota: Yes. And their actually the scale is totally different between the Japanese startups. Their variation of their IPO is a totally different, than the Japanese startup from the IPO to AAR, I think it are not good effect or whole growing their companies and the Japanese also expect such an early exit in terms of success. That is totally different.
Tim: Yeah. I agree. I mean, especially traditionally in Japan, but I don’t know, do you see that changing? I’m seeing more and more founders and I mean, Yumi, for example, that are turning down in nice safe IPO and trying to do global growth. I mean, is this something you see changing?
Kota: Yes. Yes. I feel that the variation, the more unicorns and I think there are a good example of a smart HR or network. So, I think their environment become better. Also in terms of the global startups, their number startup from startup business in the US where they have their entity in the US from.
Tim: Right, right. Like Treasure data did.
Kota: Yeah. Yeah. Treasure data. Yeah.
Tim: Well, with more and more successful Japanese role models, maybe even if the young founders don’t speak English, they’ll have enough Japanese inspiration to help them succeed.
Tim: Hey, well, listen, Kota, thanks so much for sitting down with me, man. I really appreciate it.
Kota: Yeah. Thank you very much. They were very interesting time for me.
And we’re back.
I love Ubie’s approach to UI/UX testing and design and it really highlights the importance of listening to your users, not your customers.
Hospitals, like any enterprise buyer want interfaces that mirror their understanding of the data, an interface that reflects their current forms and processes and since they’re the ones paying the bills. That’s exactly what most designers give them. And frankly, this is why most enterprise software provides a terrible user experience. It’s always simpler and safer to give the customers what they want rather than to try to tell the customers that they want the wrong thing.
Kota and his team went back to basics, talking with potential users in the park and getting feedback. You know, looking back, I suppose we should not be surprised that the final UI mirrored interfaces that the users already understood; ATM machines, karaoke machines. And it’s a wonderful example of UI design done right.
As for AI applications to symptom checking and diagnosis support, well, that is both a very interesting future and a very interesting past. In fact, one of the first practical uses for AI in the seventies and the eighties was Caduceus and similar expert systems designed to diagnose diseases or assist doctors in their diagnosis. It never quite lived up to its promise.
And even with all the advances in AI and medicine over the past 50 years, I’m not sure this generation will do much better. Medicine is science to be sure, but it’s not algorithmic or even deterministic. In many cases, even with the best information and the best minds, diagnosis and treatment involves a series of informed guesses and try and see approaches to treatment.
And to be clear, most doctors and medical startups don’t pretend that reality is any other way, but we, the public, we tend to get our hopes up.
But Ubie’s approach in particular adds a lot of value. Rather than just jumping to the final answer, they take one important direct step forward. They use AI to determine the appropriate follow up questions. It saves time and leads to better results. That’s a valuable lesson for all of us in the startup world.
Sometimes we need to put the quest for the answer aside for a moment and make absolutely sure that we’re asking the right questions.
If you want to talk more about medical AI systems or managing staff without managers, Kota, and I would love to hear from you. So, come by disruptingjapan.com/show196, and let’s talk about it. And hey, if you enjoy the show, share a link online or just, you know, tell people about it. In this age of over the top hype and reviews as a service, you’d be amazed how much power and honest recommendation has.
But most of all, thanks for listening. And thank you for letting people interested in Japanese startups know about the show.
I’m Tim Romero. And thanks for listening to disrupting Japan.