While American AI startups are dominating the headlines, one Japanese company has begun rolling out “AI employees” to famously cautious  Japanese enterprise customers.

Today we talk with Shota Nakagawa the CEO of Caster and discuss their model of human-AI collaboration, why Japan is positioned to lead real-world AI deployment, and the big steps needed for Japan to catch up with the West.

It’s a great conversation, and I think you’ll enjoy it.

Show Notes

  • Caster’s new model for gig-workers
  • Why almost 90% of Caster’s workforce are women
  • How remote work is evolving differently in Japan than in the US
  • Can remote work really revitalise rural Japan?
  • Why Caster uses full time staff rather than gig workers
  • How AI employees could be the solution to Japan’s labor shortage
  • How Caster makes extensive use of AI in their workflow today
  • What is responsoble for the low level of trust that Japanese have in AI and how to fix it
  • Which tasks AI agents will take over and which they will never do

Links from our Guest

Leave a comment

Errata

  • Caster’s percentage of female employees is about 87% not 95%.
  • Caster was founded in Tokyo, later moved to Miyazaki, and then moved back to Tokyo after the IPO

Transcript

Welcome to Disrupting Japan, Straight Talk from Japan’s most innovative founders and VCs.

I’m Tim Romero, and thanks for joining me.

While American CEOs are competing to see how quickly they can leverage AI to replace both full-time employees and gig workers alike, one Japanese company is taking a different approach and they’re already rolling out their AI workforce.

Today we sit down with Shota Nakagawa, founder and CEO of Caster. Now, Caster is a remarkably progressive and innovative Japanese company. They were a strong and vocal advocate of remote work years before the pandemic hit, and even after their IPO, their 800 person workforce remains fully remote with our corporate headquarters located in a shared office space in Tokyo.

Caster has now begun rolling out its AI workforce, and they’re taking a very Japanese approach. Rather than leveraging a collection of flexible gig workers or freelancers, Caster continues to build a long-term full-time workforce who is co-developing and already working alongside their AI employees.

If history is any guide, Caster’s thinking today might tell us what the Japanese market will be thinking 10 years from now.

Shota and I talk about the long-term AI trends in Japan, how Caster solve the problem of corporate, Japan’s deep skepticism about AI and whether or not AI can really provide a solution to the economic problems associated with Japan’s declining population.

But, you know, Shota tells that story much better than I can. So, let’s get right to the interview.

Interview

Tim: So, I’m sitting here with Shota Nakagawa, the founder and CEO of Caster, who is creator of an outsourcing platform and a relentless advocate of remote work. So thanks for sitting down with me.

Shota: Thank you very much.

Tim: So, I explained very briefly what Caster did in the intro, but I think you can explain it much better than I can. So, what does Caster do? Why is it unique?

Shota: People are all work remote every day, every day.

Tim: So, the entire company is remote.

Shota: Yeah, yeah. All people.

Tim: That’s pretty unusual. And we’re here today actually sitting in a share office space to have this conversation. Well, actually let’s talk about that in detail later. But first, let’s talk a little more about Caster. So, what do you do for your customers?

Shota: BPO, business process outsourcing, client about SMB small business, want back office service.

Tim: So, what kind of back office services, is it like recruiting, accounting?

Shota: Many types.

Tim: So, I think what Caster’s most compared to are other services like Upwork or freelancing platforms. But what you guys are doing is a little bit different.

Shota: Yeah. Most different point is trust, Caster is B2B business. Upwork is a B2C business. Trust is all different.

Tim: So, in practice that means if in the Upwork model, if something goes wrong, it’s the problem with the freelancer you hired. In the Caster model, if something goes wrong, Caster has to fix it.

Shota: Exactly.

Tim: Okay. And the other thing I’ve noticed, you use full-time staff rather than freelancers.

Shota: Yes. Yes, exactly.

Tim: So, tell me about your customers. Are they startups? Are they enterprises? Who uses Caster?

Shota: Startups, small business and lawyer, tax lawyers and social security specialists and consulting. About 30% is startup and 20% is professional lawyer, tax professional.

Tim: Okay. And tell me about your employees. If everyone is remote, where are they? What kind of people are they?

Shota: Many different. Many different. 70% are living in the local area, not Tokyo all over Japan from Hokkaido to Okinawa.

Tim: Do they fit any particular type of profile? For example, are there a lot of retirees reentering the market or there part-timers? What type of people?

Shota: 95% people is women. Very unique.

Tim: 95% are women? Why is that?

Shota: Japanese unique point. The women’s usual job is so secure. Women want a remote work job in Japan.

Tim: And is that just, are a lot of them like working mothers who want to work remotely because of children? 95% just seems really high. So, I’m curious about what the strong point is there.

Shota: The point, just could say half people have children to have lifestyle and half people local pay, so low. Caster paid to standard price attractive for the people.

Tim: I can imagine Corona changed everyone’s thinking about remote work in Japan and all over the world actually. But let’s take a step back because you were advocating for remote work a long time before Corona. From the very beginning when you founded Caster in 2014, you were talking about remote work, but what was the attitude towards remote work 10 years ago when you first started?

Shota: It’s impossible. They say it’s impossible.

Tim: Why did they think it was impossible?

Shota: Why? Because they have not tried.

Tim: Okay. That was my experience too. Corporate Japan spent 20 years saying remote work was impossible. When Corona hit in three weeks, everyone kind of figured out how to make it happen. But what was the biggest excuse companies used when they said they couldn’t do remote work?

Shota: Fear to something they don’t know.

Tim: Wow. So, it is just the unknown. Coming back to Caster and post Corona, you have around 800 staff now.

Shota: Yeah. Yeah.

Tim: Everyone’s working remotely. This is a really unusual management style in Japan. So, what did you have to change about Caster’s culture or management style to make this work in Japan?

Shota: Almost no management.

Tim: That sounds very dangerous.

Shota: Yeah. Almost no management, because remote is digital work. All actions tracking every time, but we not see all action every time.

Tim: So just measuring the results.

Shota: No, trust comes with and result.

Tim: I’ve noticed that another different thing about Caster when compared to other platforms is that you tend to assign like a project manager for each customer. So, does the project manager act as kind of that management layer?

Shota: Yes. Yes.

Tim: But what about the feeling from the workers? Do your workers ever want to get together and have that kind of social connection that’s so important in many Japanese companies?

Shota: A little, but I don’t know.

Tim: Okay. Well let’s talk about the management approach, because Caster was founded in Miyazaki. But we are here in Tokyo even though it’s a shared office, which is great. So, was it the IPO, was it interacting with investors? So it seems like there’s some aspects of business still need to be done face to face in Japan.

Shota: Yes. Good time. If they want me come.

Tim: Visit customers, is it investors?

Shota: People who have power.

Tim: So, that would be investors.

Shota: Yeah. Yeah. Investors.

Tim: But still, I think that running a public company out of a shared office is absolutely amazing. So after Corona, we’ve seen a lot of talk about remote work and some changes in attitudes. It seems that startups, small offices are adopting it very quickly, but do you think Japanese enterprises are still mistrustful of remote work, or are they embracing it more and more?

Shota: Change after Corona enterprise company decrease new hiring, difficult hiring people in Japan and now enterprise companies change the slowly but many remote work people hiring more.

Tim: So, is it just they can’t find the staff they need? Or is it that new hires want to work remotely and are demanding to work remotely?

Shota: Yeah, both of those are happening. Big company want many higher level people about marketing and professional. You can’t find them.

Tim: I think it’s interesting that recently in the US there’s a real backlash against remote work. A lot of companies are forcing people to come back to the office. But in Japan, companies, as you say, they’re slowly starting to embrace remote work more. Well, I’m going to be a little bit skeptical here. So I’m a huge fan of remote work. I’ve hired remote workers. I’ve been a remote worker. So, 70% of Caster’s workforce is outside of Tokyo. And people love to look at remote work as a way of revitalizing rural areas of Japan. And during Covid, we saw that a lot of people left Tokyo, but afterwards it seems that people have come back and that the population of Tokyo is growing again. And population of rural areas is going down again. So, is there a way to reverse the trend long term?

Shota: Do you know about 2030 population? It labor population decrease about 5 million people. About 2040 decrease by 10 million. This down cycle, there is no one who experience.

Tim: That’s true. But do you think we’ll see people move out into the country or will it just be more people coming to Tokyo and Osaka in the big cities?

Shota: To Tokyo? Yeah. Yeah. All people decrease.

Tim: Let’s talk about Caster’s business model a bit. One of the things that for a business process outsourcing remote work company that I think is unique is that all of your remote workers are full-time employees with full benefits salaries. So, why did you decide to go with full-time employment rather than a more standard freelancing model or gig work?

Shota: Few resources. We have exclusive use. Others can’t buy.

Tim: So, you have exclusive use of those resources. I guess that makes sense. So there’s a real resource shortage in Japan, whereas most of the other platforms are going after a global market where there’s a resource surplus.

Shota: Yeah. Japanese people so limited.

Tim: Well, it goes back to the population shrinking again. I think it’s actually very similar to Japan’s lifetime employment when it was started in the fifties and sixties was also because resources were very limited and they wanted to monopolize the employees for the whole lifetime.

Shota: Yes. quality. Yeah. And security.

Tim: Most of your workers, are they assigned on long-term projects or short-term projects?

Shota: 80% to 90% long project. Few years.

Tim: So, that gives you some protection, but aren’t you concerned that like an economic slowdown could leave you with a lot of unassigned staff?

Shota: But 70% is employee and 30% is freelance mixed.

Tim: Okay. So that provides a bit of a buffer.

Shota: Yeah. Yeah.

Tim: Excellent. Let’s talk about AI Just a few months ago you announced a really interesting new joint venture and pilot project for AI workers. Can you tell me a little bit about that?

Shota: Okay, I want to make a cyborg. Cyborg is mixed AI and human resource.

Tim: Okay. The new joint venture with Alt.AI, you’re creating AI assistance and agents and AI employees on the Caster platform. What kind of tasks are they going to do?

Shota: Main tasks is communication, many communication jobs in the works. I think all jobs, 70% communication, all people, all jobs. Communication is special to AI.

Tim: So, so what kind of communication? Are they writing reports? Are they building PowerPoint presentations? Are they sending email? What are they doing?

Shota: They do that and chat, call the bosses and interact their staff members. All jobs. Communication is — this meeting is communications. Communications is many types.

Tim: Can you give me a case study, like an example? Walk me through how one of your customers might use one of these AI agents or AI employees for better communication.

Shota: If you order a task, a job to be done; first communication chat, you order your job. Next, our manager, the assigned employee and freelance build team. This process is all communication.

Tim: So, is the AI making the assignments?

Shota: Yeah. Making assignments and client communications.

Tim: So, it’s doing a lot of the functions of a project manager or a manager.

Shota: Yeah.

Tim: So what’s been the usage and feedback from the customer so far?

Shota: They don’t notice AI, AI is hidden.

Tim: It’s just a smoother communication.

Shota: Yes. And like a human, don’t notice.

Tim: That’s really interesting because, well, overall, how much trust do you think Japanese business and Japanese people have an AI?

Shota: No percent. Yeah.

Tim: They don’t trust it yet.

Shota: Yeah. But human jobs are 100%. We provide remote work resource. Remote work resource is digital, team AI resource. So, AIs provide new resource to client. Don’t notice client.

Tim: That makes sense. So, far you’re saying that the AIs are mostly involved with like improved communication, making assignments. Do you think they could be used in like accounting or legal or things that has to be very precise?

Shota: Yeah, exactly. Human have mistake. AI is a not have mistake.

Tim: Yeah, that’s right. Humans make mistakes too. That’s a good point. I just find like humans are more trainable. Humans tend not to make the same mistake more than once. AIs will keep making the same mistake.

Shota: I don’t think so. We can train AI and not do a same mistake. AI can’t lie.

Tim: Well, I think that’s really the key of getting that broad acceptance. So, what’s the next target? Are you looking at like AI agents for programming tasks or Excel or what? Outside of communication, what do you see as the most exciting area that AI agents can do in Japan?

Shota: In general, communication, management and operation. Communication is 70% all tasks and 20% management, 10% is operation. So, few operations, communication is a main business area. AI next management.

Tim: To turn it around, the areas that AI will not be able to solve are on that 10% of operations.

Shota: Yeah, exactly. AI is not take this cup.

Tim: Right. It can’t physically move things around. How extensively are you using AI internally, for example, for your project managers or for business process outsourcing?

Shota: Now, so few, but one years, two years in future half 50% AI do.

Tim: Yeah. 50% of operations being done by AI.

Shota: Yeah. Communication management operation, 50% over.

Tim: So, do you see AI as the potential solution to Japan’s declining birth rate and declining population?

Shota: Exactly. Japan have big chance people, a total chance if a population is growing and not chance population is decreasing. Yes. Business chance.

Tim: Yeah, that makes sense. It’s innovation only really comes when there’s some pain being felt. And in markets like America where there’s lots of inexpensive labor or other Southeast Asian markets where labor is cheaper, it’s not as urgent to deploy AI. But in Japan, there’s a huge need for any kind of worker.

Shota: Yeah, yeah, exactly.

Tim: Well, listen, Shota, before we wrap up, I want to ask you what I call my magic wand question. So, 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, whether that’s the attitude towards risk, the education system, anything at all to make it better for startups and innovation in Japan, what would you change?

Shota: Change the democracy.

Tim: That’s interesting. How would you change it?

Shota: Change Imperials. Go imperials.

Tim: How would that spur innovation? How would that make things more innovative?

Shota: Rules are too strict in Japan now.

Tim: So having a strong leader who could just change the rules, would spur innovation?

Shota: Break the rule.

Tim: Well, I’ve got to say that’s Donald Trump’s whole theory right now, right?

Shota: Yeah. Yeah.

Tim: Break the rules, simplify anything to create more innovation. All right. I don’t see that happening in Japan, but it’s a magic wand. But do you think the rules are getting looser or are they still just too tight in Japan?

Shota: No.

Tim: Still far too tight.

Shota: Yeah. Japanese people is so polite, so polite.

Tim: I think that’s true. Japanese people respect rules and they feel comfortable with…

Shota: Yeah, yeah.

Tim: It’s very interesting. In English, we talk about following the rules or obeying the rules, but in Japanese it’s Mamoru. You’re protecting the rules.

Shota: Yeah, protecting.

Tim: It’s a very different way of looking at regulation, but hopefully we’ll see that change sometime in the future to get more innovation. Well, listen, Shota, thank you so much.

Shota: Thank you. Thank you so much.

Outtro

And we’re back.

Shota’s approach to AI is well aligned with the overall sentiment in the Japanese market as a high wage, high skill, scarce labor market, Japan is an ideal place to deploy practical AI applications.

But, you know, there are cultural reasons as well as economic ones. Japanese labor has always been very accepting, if not outright supportive of automation and robotics. American labor, on the other hand, tends to be outright hostile towards it.

CEOs of Japanese robotic startups have told me that during pilot programs, they were shocked to find that while Japanese workers would actively participate and offer ideas for improvement, American workers would refuse to provide feedback and sometimes actually sabotage the equipment.

Of course, this American worker behavior is not necessarily based on ignorance. A lot of it is rational self-interest. In Japan, there’s an expectation that the employer will use the savings to expand the business and find alternative work for those whose jobs is eliminated by automation. It is both legally and socially very difficult to fire people in Japan.

As Shota pointed out, Caster’s customers prefer the stability, consistency, and quality that comes with long-term employees over the low cost, or perhaps I should say, low short-term costs of gig based outsourcing platforms.

The US on the other hand, fosters an adversarial relationship between labor and ownership. CEOs proclaim publicly and excitedly that AI will allow them to cut their existing workforce and increase profit margins. Billboards in San Francisco proudly proclaim, “Stop Hiring Humans!”

So yeah. That part of the AI backlash. It makes a lot of sense.

That said, however, innovation is moving much faster in the US and AI is deploying much more quickly there. A recent nationwide survey by Nomura Research Institute showed that only 9% of Japanese have used generative AI while similar research from Harvard showed that 40% of Americans have.

So, how do we really drive innovation in Japan?

Well, Shota’s comment about bringing back the emperor system is not as crazy as it sounds. I mean, yes, obviously it’s completely crazy on the face of it, but he gave us an important insight here.

Innovation requires constraints to push against.

Some of the most iconic Japanese art and theater was created by pushing against the strict constraints of the Eldora laws. Japan’s incredible technological innovations during the post-war period were driven largely by tight capital constraints. And the small size of the domestic market.

Startup innovation, at least traditionally, has been driven by startups, lack of capital and existing customer base. The ticking clock provided the motivation for innovation and the small size gives startups the agility needed to execute.

Interestingly, however, the current crop of AI startups are doing something different. They come into being with billions of dollars in capital and large corporate contracts already in place. I mean, last year, Safe Super Intelligence raised $1 billion at a $5 billion valuation in their pre-seed round. No product, no prototype, just three really smart guys.

It’s hard to describe just how bonkers this is from a startup point of view. A $5 billion valuation is an extremely successful exit, not a pre-seed valuation. These companies aren’t really startups. They’re something else entirely. It’s a new model, but they’re doing cool things and I wish them the best.

If you want to see real startup innovation in action, look at Deep Seek. When faced with severe constraints in both hardware and capital, they innovated and they came up with an innovative, low cost, more practical solution.

That’s how startup innovation is supposed to work.

In Japan, we have the constraints. We just need the motivation to push harder against them.

 

 

If you want to talk more about delivering AI innovation in Japan, Shota and I would love to hear from you. So come by disruptingjapan.com/show232, and let’s talk about it. And hey, if you enjoy disrupting Japan, share a link online or just, you know, tell people about it. Disrupting Japan is free forever and letting people know about it is the absolute best way you can support the podcast.

But most of all, thanks for listening. And thank you for letting people interested in Japanese startups and VCs know about the show.

I’m Tim Romero and thanks for listening to Disrupting Japan.