【amoibe Co., Ltd.】Solving Engineer Shortage with "Virtual OJT" and AI Mentors. Digital Job Training "amoibe OJT" Driving AI Shift and In-house Development in the SI Industry

Structural Reform in the SI Industry Starting with the Rise of AI. Breaking the "Project Gacha" to Revitalize Engineers

Thank you very much, everyone. My name is Hayato Shinjo, Representative Director of amoibe Co., Ltd.

We are a generative AI x education technology startup tackling the theme of the "AI shift in the SI (System Integration) industry."

Currently, while the Japanese SI industry boasts an extremely large market scale, it maintains a complex, uniquely Japanese industry structure like multi-tier subcontracting and on-site dispatch. However, with the rise of AI-driven development tools represented by Cursor and Devin, a disruptive structural reform is about to take place.

The most prominent manifestation is the "polarization" of the engineer shortage in the field. While highly skilled engineers who can utilize AI are in critical shortage, conventional developers who can only do simple coding are starting to become redundant.

Furthermore, due to the structure of the SI industry, there is a fatal challenge in that it is very difficult to "train engineers with new technologies through actual work in the field." In the industry, the term "project gacha" is often used. If a project to which an engineer is assigned happens to use legacy (old) technologies, they have no opportunity to touch latest cloud technologies or modern frameworks, regardless of their own excellence.

Therefore, we developed "amoibe OJT," a digital platform that completely replaces corporate OJT (On-the-Job Training). OJT is an old mechanism that has hardly changed since World War I, but we have redefined it into a completely new training method for the AI era.

Zero Senior Staff Labor. Handling Tasks in Virtual Projects and Receiving Individual AI Reviews on QCD and Technology

"amoibe OJT" provides training engineers with a mock "development project" constructed in a virtual environment on the cloud.

For example, they join a project like "inventory management system development" and handle incoming tasks and tickets while writing actual source code.

In this system, it is the "AI Mentor" that accompanies them instead of conventional superiors or senior employees. The AI Mentor does not act as a simple chatbot but behaves like a real team leader. It demands "Ho-Ren-So" (Report, Contact, Consult) from trainees, and checks the submitted code from the perspective of "QCD" (Quality, Cost, Delivery), hard skills (technical accuracy), and soft skills such as logical thinking and communication, providing individual and super-concrete feedback (reviews).

Unlike lecture-style or task-clearing learning at conventional e-learning or general programming schools, we train engineers in a "practical environment" that is extremely close to actual work. Moreover, the biggest strength is that in the process of instruction and code review, which takes the most effort, it "does not consume any labor (time cost)" of the company's excellent senior engineers or superiors.

In companies with many engineers, the system is designed so that the more training data accumulated, the more the AI Mentor self-learns and grows into a "more excellent and instructive senior employee," creating a positive loop.

Dramatic results have already been achieved in many companies. In an introduction case of a major IT engineer dispatch company, we had engineers who were hired with no or little practical experience and were on standby (red ink for the company) because of a lack of skills, undergo training for 15 business days. As a result, they became immediately effective in just half a month, and all were assigned to field projects by the end of the month, achieving a turn to black ink at once. The fact that the ROI (return on investment) can be clearly measured is a major strength of our product.

150% Monthly Growth and Over 400% NRR. From Training to Advanced Database Construction, and Entry into the Core of SI

Our business model is a simple design of a "metered rate system" from 300,000 yen per trainee.

In a case of a major SIer's Nagoya branch, it started from a mock PoC (proof of concept) of 3 people at first, but following the remarkable growth of the engineers, it expanded to nationwide development the following year. 45 people took the course, growing into a major account exceeding 15 million yen annually combined with other development courses.

Due to this high cost-effectiveness, sales are growing rapidly at a pace of over 150% monthly. Especially noteworthy is the large amount of "repeat contracts," and the "NRR" (Net Revenue Retention), an index showing the sales maintenance and expansion of existing customers, exceeds "400%." This means that if we establish 100 million yen in sales this year, sales will stack up to 400 million yen next year purely through repeats and upsells of existing customers without any new sales activities.

As a roadmap for the future, we will open this training business to individuals (consumers) as well, constructing an "AI-native advanced engineer database" that can perfectly use AI-driven development tools (Cursor, Devin, etc.) by ourselves. So to speak, it is a platform that purchases low-value "water," refines it, and sends it out as high-value "oil (advanced human resources)."

And our ultimate goal is not to end as a simple "training company." Armed with this powerful "high-speed training device" and "advanced human resources database," amoibe itself will directly enter the SI (system development) market.

The current Japanese SI industry consists of a multi-tier subcontracting structure with man-month unit pricing, optimized to the needs of large companies that do not have in-house development functions. However, in the coming era when "making" itself is commoditized by AI, large companies will inevitably shift to "in-house development." We do not aim to compete with other companies for outsourced development slots, but rather aim to become a new infrastructure of the SI industry, supporting the shift to in-house development of Japanese companies through the provision of human resources databases and training platforms.

We have set a sales goal of 10 billion yen in 2030. I founded a startup in 2014 at the age of 24 and experienced business sale, but it was not a huge success, and I have continued to ask myself strong questions about not leaving an impact on society. Under the strong passion of "changing myself and supporting social reform," I lead this business with a team of industry experts. We ask for alliances and support from all of you to strongly promote the AI shift in the SI industry together. Thank you very much.

Q&A and Feedback

Commentator (Mr. Ito): Thank you very much, Mr. Shinjo, for your passionate and excellent presentation. Having run an SIer myself in the past, I understand the deep-rooted nature of the multi-tier subcontracting structure, the dispatch man-month model of the Japanese SI industry, and the cost burden of running OJT in the field all too well.

As a question: with the future AI shift, how do you think conventional "man-month pricing, labor-intensive" SIers will change, and how is amoibe planning to cut into the market specifically? Please tell us including trends such as the transition from waterfall to agile.

Mr. Shinjo: Thank you for the question.

It is a great honor to receive a question with such high industry resolution from Mr. Ito, who knows the industry deeply.

We do not think of a radical scenario where AI will wipe out all existing SIers tomorrow. We believe that the future market will be divided into roughly three gradations.

One is the area of legacy systems using old mainframes or COBOL. Since it is difficult to load generative AI there in the first place, the conventional multi-tier subcontracting and dispatch man-month model will continue to remain at a certain ratio.

Another is the cutting-edge area where DX departments of large companies completely steer towards "vendor departure (in-house development)" and complete agile development solely with in-house engineers who can use AI-driven development tools. Outsourced vendors are unnecessary there.

And the largest market is the "gradation area that continues to outsource but has to refresh the contents into AI processes" in the middle. If development by AI agents progresses, productivity will jump by 5 or 10 times, so the business model of determining dispatch unit prices by "man-months" will collapse.

To adapt to this change, we provide "OJT platforms to support in-house development" to client companies and "high-speed retraining systems to adapt to the AI era" to IT companies on both sides. Rather than becoming a simple competitor for outsourced development, our clear strategy is to establish a position functioning as the "infrastructure of the AI shift" for the entire SI industry.

Mr. Ito: I see. Supporting the transition to a "productivity-oriented development process" after the collapse of the man-month model through retraining on both sides, and reigning as infrastructure. I think it is a very rational strategy.

Another point: I felt that this model of "running OJT with AI in a virtual environment" is a very powerful education framework that can be expanded horizontally to various industries and occupations, not just SIers or the IT industry.

For example, with the increase in telecommuting (working from home), I often hear cries in trading companies and general enterprises that "OJT for young employees is not functioning and people are not growing." Furthermore, since the concept of OJT is not originally developed in foreign countries as in Japan, I think the demand for packaged OJT training is large. What do you think about the possibility of this horizontal expansion?

Mr. Shinjo: That point is precisely an important approach to expand the possibilities of our business greatly.

As you pointed out, our system is essentially a "digital job training (highly advanced job training that can only be reproduced in a virtual space)." For example, it is the same essence as a doctor performing a vascular bypass surgery simulation in a VR space before surgery, or an astronaut performing rover operation training in a simulator.

This mechanism of "business process simulation and real-time feedback by AI" can be expanded horizontally to other occupations, such as role-playing for sales positions, response training for customer support, and trading operations for trading companies.

At present, because the market pain and business opportunities of the engineer shortage and AI shift in the IT/SI industry are so large, we are concentrating our management resources on the engineer area first to aim for the top share. However, after establishing the trademark and mechanism of "digital OJT" there, we have a very realistic roadmap to expand to other occupations and globally.

Mr. Ito: Building an overwhelming vertical position in the deep pain of the engineer area first, and then expanding horizontally from there. I was convinced that it is a business with high expectations, including the prioritization as an entrepreneur. I hope you will succeed for the sake of raising Japan's digital technical capability. Thank you very much.

Mr. Shinjo: Thank you very much.