Switch to the Japanese section - View the Japanese version here
In 2026, Sakana AI released a series of products: Sakana Chat, Sakana Marlin, Sakana Fugu, and Sakana Translate. Known until now as an R&D company, Sakana AI has entered the stage of delivering products born from its research to companies and government agencies. The Go-To-Market (GTM) team that sells and scales those products launched in earnest in September 2026.
Leading the team is Hidekazu Kakishima, who has driven the business launches of new technologies like cloud, IoT, and drones right from the dawn of their respective industries. Why Sakana AI, and why now? How do you sell products that come out of research, and what kind of team do you build to grow them? And what makes it interesting to join a team that still has few established systems? We sat down with him to find out.
Interviewee
 |
Hidekazu Kakishima VP of GTM Led the strategic accounts team during the early days of AWS Japan, then joined IoT connectivity startup SORACOM as a founding member, and later drove the real-world adoption of flying robotics at US drone maker Skydio. He now leads market expansion and value delivery for Sakana AI’s product lineup. |
The Frontier of Technology Draws Talented People Who Love What’s New
— From AWS to SORACOM, Skydio, and now Sakana AI, you seem to have chosen the early days of a new technology every time. Tell us about your career.
At the root is a belief that technology makes the world better. That is why I have always wanted to work at the leading edge of fields where new technology is emerging. I was not that conscious of it in my twenties, but once I encountered cloud computing at AWS, I switched firmly into that mode.
After that came IoT at SORACOM, drones at Skydio, and now AI. Each is a field where a new industry is being born, and my job is to take the technology at the frontier and put it to work in the market from the inside. That is what I want to do, and I have chosen every step of my career by that standard.
There is one more reason I keep choosing the launch phase. When a new trend emerges, its leading edge attracts talented people who love new things. Learning the latest technology, implementing it, and seeing customers happy is enjoyable in itself. But working alongside those people is the particular pleasure of the early days.
— Given that, why Sakana AI?
Because I thought it was the best AI company in Japan. It is growing fast as a startup, its valuation is rising rapidly, and it is a Japan-born company aiming at the global market with research and development as its foundation. I had been following the company for some time. Then an acquaintance introduced me, and I had the chance to actually hear the story.
Before joining, I spoke with Ren Ito, co-founder and chairman. Hearing his vision of scaling Sakana AI as a business not just in Japan but globally was exciting. The founders and the members are all wonderful people, and in the end, the desire to work with good people was what decided it.
Because the company started from R&D, it has excellent technology and attracts talented people. But turning that into something the world can use is not easy. I felt the company was building a team for exactly that and expanding rapidly, and the GTM team now owns that part: making it usable.
From Research Insights to Products That Reach More Customers
— Let’s talk about that mission of making the technology usable. Where does the GTM team stand today?
Alongside its research, Sakana AI has focused on industries such as finance and defense and implemented AI on the ground at customers. We have now entered a phase of turning the knowledge accumulated there into products and delivering them to a broader range of customers.
That said, releasing a product does not mean customers can use it right away. There will always be situations where the product alone cannot meet a customer’s requirements. To close that gap, we are currently hiring Forward Deployed Engineers (FDEs).
FDEs feed what they learn at the customer’s site back into the product. As that accumulates, we want to increase the number of cases where the product can be used on its own, without an FDE.
Many AI companies are investing in the FDE role right now. At the same time, people sometimes ask how it differs from an engineer who goes on site and builds things for the customer. The difference is precisely this feedback loop. Solving the customer’s problem is not the end. Reflecting that experience back into product improvement is part of the job. That is the role of an FDE at Sakana AI and the model we are aiming for.
The Product Opens the Conversation; the Customer’s Problem Is What We Solve
— The job descriptions for the GTM team use the phrase “product-driven GTM.” It sounds different from both system-integrator development and deploying a SaaS product as is. What is the approach?
When people hear “product-driven,” many picture a one-way push to sell a product that already exists. Our stance is a little different. For us, product-driven means using the product as the starting point for a conversation with the customer. We present the product first, watch the customer’s reaction, and then propose the best way to apply it as a solution. It is less about going out to sell a product and more about matching the product to the customer’s problem.
There are also cases where the customer does not know how to implement it, or where security has to be tailored to specific requirements. These requirements differ by industry and by company, so an FDE steps in to move the deployment forward.
When there are no special requirements, customers can use the product as is. Our API services, for example, are called directly from the customer’s own systems. Sakana Marlin is a research product for business users, aimed mainly at people who do research work every day.
— We understand you are not selling only through your own team, but also place importance on working with partners.
Our resources are limited, so partners are essential. One type is system integrators, whom we are encouraging to build Sakana AI products into their own solutions. The other is technology companies with partner ecosystems, where Sakana AI products are offered through their marketplaces.
This model lets us grow the user base not just in Japan but globally.
Orchestrating Multiple Models, and the Growing Demand for AI Sovereignty
— A bit more on products. Orchestration, which brings in multiple frontier models, is one of Sakana AI’s technical strengths. How are customers responding?
Many companies have recently released features that let you switch between different models. Sakana Fugu does not simply switch models, though. It takes the approach of having multiple frontier models work together on a single task. I think that is a very distinctive feature.
Behind the demand for this approach is a wariness about continuing to depend on a single model, however good it is. In February 2026, the US government instructed federal agencies to stop using Anthropic products, and the US Department of Defense designated the company a supply chain risk. No matter how strong the performance, a model can become unavailable for political or contractual reasons. Many companies must have felt that firsthand.
More and more companies want to use multiple models, and this approach is drawing attention. We also hear concerns that a particular model could suddenly become unavailable. For better or worse, these circumstances are a tailwind, and we are receiving inquiries from many customers.
— Are you seeing changes in customer attitudes around AI sovereignty as well?
I think so. AI is a bigger trend than anything before it, and precisely because of that, awareness of security is extremely high. Awareness of AI sovereignty and data sovereignty is rising among both customers themselves and governments, and requests of this kind are clearly increasing.
And it is not limited to large companies. Small and medium-sized businesses are also interested from a data sovereignty perspective. It has become a theme shared across a wide range of customers.
— Demand is high, yet both customers and sellers still seem to be feeling their way. Compared with cloud and IoT, which you have worked on before, what is the same and what is different?
First, the AI market is overwhelmingly larger. Beyond that, the customer’s situation is different. Cloud had an easy-to-understand framing as a replacement for on-premises. IoT had M2M communications before it, so customers understood it quickly, and the decision was often simply to migrate if it improved efficiency.
AI is not like that. The customers themselves are feeling their way. They do not know what to do. And because information from around the world now arrives flat over the internet, AI players and customers are struggling at the same time. That makes the job of selling harder.
Still, customers have a strong sense that they have to do something. We can approach them exactly where they do not know how, so the business is moving at tremendous speed. On the other hand, models are evolving so quickly that customers tend to avoid long-term commitments. They also want an environment they can use with confidence and safety.
— In that market, where do you aim to be in one year and in three?
The immediate priority is to build up effective deployment results with customers in Japan. But when it comes to expanding globally, waiting until we have a track record in Japan and then taking it overseas would be too slow. So we are moving with the intention of pursuing Japan and overseas in parallel from the start.
In Japan, we will expand broadly through collaboration with the investors who have backed us and through partnerships with a range of partners. Overseas, the thinking is the same: build results with the overseas big tech companies and other investors who have backed us, and scale from there. Ideally, in three years, overseas will account for the larger share.
We Want People Who Enjoy an Environment That Changes Fast
— Let’s turn to the team. The GTM team is hiring for several roles: Account Executive, FDE, Partner Alliances, and Customer Success. For each role, what kind of person do you want to join?
Let me start with what applies to every role. One is enjoying change. Both the industry and the company are moving very dramatically right now, and we want people who can enjoy that and thrive in it. The other is curiosity. Building your work while engaging with a wide range of people, including customers, the industry, and colleagues inside the company, requires curiosity.
On top of that, for Account Executives, the market is large and we expect the deal sizes we sell to grow dynamically. We are looking for people who have built large businesses from zero and closed large deals. Partner Alliances handles the collaboration with system integrators and ecosystems I mentioned earlier. It is a job of designing from scratch who we partner with and how we sell together.
— What kind of person are you looking for as an FDE?
For FDEs, I do not want them to start from “we can’t” when a customer makes a request. The role is to look at the customer’s requirements against our capabilities from a technical standpoint and close the gap between them. Someone who can carry a problem through from discovery to resolution on the technical side is a good fit.
Beyond that, we want people who can look broadly for the next problem and the next opportunity, even outside their scope. Do not stop at solving the issue. Find what is needed next and connect it to product improvement. That is an FDE at Sakana AI.
— Customer Success, a new role you are hiring for, handles customers after deployment, correct?
Yes, it handles follow-up after a customer has deployed. Because models evolve quickly, we need to keep checking whether customers are using them properly and whether new issues have emerged. We will be releasing products aggressively from here on, so combining them into a more efficient setup for the customer is also part of Customer Success’s job.
By the nature of the role, Account Executives move on to the next customer. If deployed customers are left alone, they may switch to another company or simply lose track of what to do. Supporting them there is a very important role.
Speaking at a Datadog event, October 2026
Now Is the Only Time You Can Build the System With Us
— What can you gain by joining the GTM team now?
There is still a great deal that has not been systematized, so you get to build those systems with us. That is the biggest thing.
Members from a range of backgrounds are joining, each bringing their own ways of thinking and opinions to the discussion. That produces new insights, and moments where you realize what it means to build something new. Opportunities for this kind of experience are rare. Building from zero at a company riding the big wave of AI is a valuable experience.
And then, from Japan to the world. This is a challenge only Sakana AI can offer. Sakana AI’s development base is in Tokyo. The researchers and the product development team sit at the desks right next to the GTM team. Feedback and feature requests from customers can be discussed with the development side on the spot and reflected in the product right away. There is no waiting for a reply after raising a request to an overseas headquarters.
We want to deliver technology and products born in Japan to customers around the world at this same speed. We are waiting for people who will build that with us.
Careers
Delivering Japan-born AI products to customers around the world. The work of building the system to do that has only just begun. If you are interested in Sakana AI’s GTM team, please visit our careers page.
In 2026, Sakana AI released a series of products, including Sakana Chat, Sakana Marlin, Sakana Fugu, and Sakana Translate. Known until now as a research and development company, Sakana AI has entered the stage of delivering products born from its research to companies and government agencies. The GTM (Go-To-Market) team responsible for selling and scaling them was formally launched in September 2026.
Leading the team is Kakishima, who has driven business launches since the dawn of new industries formed by new technologies such as cloud, IoT, and drones. Why Sakana AI, and why now? How do you sell products born from research, and with what kind of team do you expand them? We also asked him where the appeal lies in joining this team right now, while its structures are still being built.
Interviewee
|
|
Hidekazu Kakishima Hidekazu Kakishima VP of GTM He led the strategic accounts team during the founding period of AWS's Japan subsidiary, then became a founding member of IoT connectivity startup SORACOM, and at U.S. drone maker Skydio promoted the real-world deployment of flying robotics. He is currently responsible at Sakana AI for the market rollout and value delivery of its product lineup. |
At the forefront of technology gather excellent people who love new things
-- Looking at your career, from AWS to SORACOM, Skydio, and now Sakana AI, it seems you have consistently chosen the launch phase of new technologies. Could you tell us about your career so far?
At the root of it is the belief that technology makes the world better. That's why I've always wanted to work at the forefront of fields where new technologies are being born. I wasn't that conscious of it in my twenties, but once I encountered AWS and cloud computing technology, I think I clearly entered that mode.
After that, I worked on IoT at SORACOM and drones at Skydio, and now it's AI. In each case, these are fields where new industries are being born, and from a position within them, I implement cutting-edge technology in the market. That's what I want to do, and I've always chosen my career by that standard.
There is one more reason I have kept choosing the launch phase of new fields. When a new trend emerges, excellent people who love new things gather at its forefront. Learning the latest technology, implementing it, and making customers happy is of course enjoyable in itself, but I believe the real thrill unique to a launch phase is being able to work alongside such people.
-- With that in mind, why Sakana AI?
Because I thought it was the best AI company in Japan. It is growing fast as a startup, and its corporate value is rising rapidly. It is also a company that started in Japan, aims globally, and has research and development as its foundation. I had been paying attention to it for some time, but an acquaintance introduced me and I got the chance to actually talk with them.
The person I spoke with before joining was co-founder and president Ren Ito. When I heard his vision of growing Sakana AI as a scalable business not only in Japan but globally, I was extremely excited. The founders and members are all wonderful people, and my desire to 'work with good people' was the final deciding factor.
Since the company began with research and development, it has excellent technology and gathers outstanding people. However, turning that into something usable around the world is by no means easy. I felt that the team was being built toward that goal and that the company was in the midst of rapid expansion, and the GTM team is now precisely the part responsible for 'making it usable.'
Delivering knowledge cultivated through research broadly as products
-- Let me ask about the mission of 'making it usable.' What is the situation of the GTM team now?
In addition to research and development, Sakana AI has focused on industries such as finance and defense and has implemented AI at customer sites. We are now entering a phase where we turn the knowledge accumulated there into products and deliver them to a broader range of customers.
That said, it is not so simple that customers will start using a product as soon as it is released. There are inevitably situations where a product alone cannot fully meet a customer's requirements. To bridge that gap, we are currently hiring FDEs (Forward Deployed Engineers).
FDEs feed back the knowledge gained at customer sites into the products. By accumulating that, we want to increase the number of products that can be used on their own without an FDE in the future.
FDE is a role that many AI companies are focusing on. On the other hand, we are sometimes asked, 'How is this different from engineers who go into customer sites and develop there?' The difference between such engineers and FDEs lies precisely in this feedback aspect. The role and ideal of Sakana AI's FDE is not to finish once a customer's problem is solved, but to reflect that experience in product improvement.
The trigger is the product, but what we solve is the customer's problem
-- The description of the GTM team's open positions uses the phrase 'product-driven GTM.' It feels different both from development by system integrators and from simply introducing a SaaS product as-is. What kind of approach is it?
When people hear "product-driven," many probably imagine a one-sided sales approach that starts with the product. But our stance is a little different. What we mean by product-driven is using the product as a trigger for dialogue with customers. First we present the product, observe the customer's reaction, and then propose the optimal way to use it as a solution. Rather than going out to sell the product, it feels closer to applying the product to the customer's challenges.
There are also situations where customers don't know how to implement it, or where security needs to be tailored to individual requirements. These requirements differ by industry and by company, so FDEs step in there and drive adoption forward.
On the other hand, if there are no individual requirements, customers can use the product as-is. For example, our API service is used by calling it directly from the customer's system. Sakana Marlin is a research product for the business layer, and its main target is people who work on research tasks every day.
-- I understand that you don't sell only with your own team, but also place importance on collaboration with partners.
Our resources are limited, so we can't do without the power of partners. One type is system integrators (SIers), and we are encouraging SIers to incorporate Sakana AI's products into their respective solutions. The other is technology companies that have partner ecosystems. In such companies' marketplaces, we have them offer Sakana AI's products.
With this model, we can increase the number of users who can use our products not only in Japan but globally.
"Orchestration" that coordinates multiple models, and the response to AI sovereignty
-- Let me ask a little more about the products. I believe "orchestration," which incorporates multiple frontier models, is one of Sakana AI's strengths. How are customers responding?
Many companies have recently released features that let users use different models selectively. However, Sakana Fugu takes an approach not of simply switching models, but of coordinating multiple frontier models to work on a single task. I think this is a very unique characteristic.
Behind the demand for this approach is a wariness about continuing to depend on a single model, no matter how excellent it may be. In February 2026, the U.S. government instructed federal agencies to stop using Anthropic products, and the U.S. Department of Defense designated the company as a supply chain risk. No matter how high the performance, a model can become unusable due to political or contractual circumstances. Many companies must have felt that firsthand.
An increasing number of companies want to use multiple models selectively, and this approach is attracting attention. I also hear concerns that a specific model might suddenly become unavailable. These circumstances, for better or worse, are acting as a tailwind, and we are receiving inquiries from many customers.
-- In the context of AI sovereignty as well, are customer responses changing?
I feel they are changing. AI is a trend larger than anything before, and precisely because of that, awareness of security is extremely high. Awareness of "AI sovereignty" and "data sovereignty" is rising among both customers themselves and governments, and such requests are clearly increasing.
Moreover, it is not limited to large enterprises. Small and medium-sized companies are also interested from the perspective of data sovereignty. It has become a theme common to a wide range of customers.
-- While demand is high, I feel that both customers and sellers are still feeling their way. Compared with technologies you have been involved with before, such as cloud and IoT, what is the same and what is different?
First, the AI market is overwhelmingly large. On top of that, the state of customers is different. With cloud, there was a clear-cut picture of replacing on-premises systems, and with IoT, M2M communication had existed for some time, so customers understood quickly, and many decisions were made on the basis of whether it would improve efficiency, in which case they would migrate.
AI is not like that; customers themselves are feeling their way. They don't know what to do. Moreover, because information from around the world now reaches everyone flatly via the internet, both AI players and customers are struggling at the same time. As a result, the difficulty of sales has increased.
However, customers have a strong sense of urgency that they "have to do something." Since we can approach them where they don't know how, the business is moving with tremendous momentum. On the other hand, models are evolving so fast that customers tend to avoid long-term commitments, and in addition, they seek an environment where they can use AI safely and securely.
-- In such a market, where are you aiming one year and three years from now?
Our immediate priority is to increase effective adoption track records with customers in Japan. However, when expanding business globally, if we try to take it overseas only after building a certain track record in Japan, it will be too late. So we are operating with the intention of proceeding in Japan and overseas in parallel from the start.
In Japan, we will expand broadly through collaboration with the investors who have funded us and through partnerships with various partners. Overseas, we take the same approach: we will build a track record with the overseas big tech companies that have invested in us and then scale from there. Ideally, three years from now, the overseas share will be larger.
What we want are people who can enjoy an environment of drastic change
-- From here, I'd like to ask about the team. The GTM team is hiring for multiple roles: Account Executive, FDE, Partner Alliance, and Customer Success. What kind of people do you want for each role?
First, let me start with what applies to all roles. One is the ability to enjoy change. Right now, both the industry and the company are moving very drastically, so we want people who can enjoy that and thrive. And the other is curiosity. Curiosity is essential for building work while engaging with a wide range of people, including customers, the industry, and internal members.
On top of that, for the Account Executive role, we expect the market to be large and the amounts we sell to grow dynamically. We would prefer people who have built large businesses from scratch or who have closed large deals. Partner Alliance is the role responsible for collaboration with the SIers and ecosystem I mentioned earlier, and the job is to design from scratch who to partner with and how to sell together.
-- What kind of people are you looking for as FDEs?
For FDEs, I want them not to start with "we can't do that" in response to customer requests. From a technical standpoint, the role is to fill the gap when comparing customer requirements with our capabilities. It suits people who can see things through on the technical side, from discovering issues to resolving them.
Furthermore, I want people who can broadly identify the next issues and look for opportunities even outside the scope. Rather than stopping once something is solved, they identify what is needed next and connect it to product improvement. That is what a Sakana AI FDE is.
-- The newly recruited Customer Success role is responsible for customers after implementation, right?
Yes, it handles follow-up after implementation. Since models evolve quickly, we need to continuously check whether customers are using them properly and whether new issues have arisen. We will be actively releasing products going forward, so combining them to create more efficient systems is also part of Customer Success's job.
By the nature of the role, Account Executives move on to the next customer. On the other hand, if implemented customers are left alone, they may move to another company or simply become unsure of what to do. I believe the role that supports them there is very important.
October 2026, speaking at a Datadog event
Only now can you build the systems together
-- What can you gain now by joining the GTM team?
There are many areas that have not yet been systematized, so being able to build those systems together. I think that is the best part.
Members from various backgrounds are joining, and we bring together our respective ideas and opinions for discussion. From there, new insights emerge, and you can really feel that "this is what it means to create something new." Opportunities to have this kind of experience are rare. Creating something from scratch at a company in the middle of the huge wave of AI. I think this is a valuable experience.
And then, from Japan to the world. This is a challenge that only Sakana AI can take on. Sakana AI's development base is in Tokyo. Researchers and the product development team sit at desks right next to the GTM team. Feedback and feature requests from customers can be discussed on the spot with the development side and immediately reflected in the product. There is no need to send requests to an overseas headquarters and wait for a reply.
We will deliver technology and products born in Japan to customers around the world at this same speed. We look forward to welcoming people who can build that together with us.
Careers
Delivering AI products born in Japan to customers around the world. The work of building the systems for that has only just begun. If you are interested in Sakana AI's GTM team, please take a look at ourcareers page.