Building an Organization Ready to Take on AI Projects: How AI Engineer Staffing Strengthened Development Capabilities

Case Study | AES (AI Engineer Staffing)

TechFirm Inc. is primarily engaged in contract system development. Its DX Design & Consulting Department was established based on the belief that simply building a system cannot produce the best outcome when the client has not yet organized its upstream requirements. The department therefore focuses on upstream consulting and overall DX planning.

While inquiries related to AI utilization were rapidly increasing, the department faced a challenge: it could not consistently secure enough engineers capable of implementing AI into clients’ services. With this imbalance between demand and available resources, even when new opportunities emerged, the team was not always able to launch projects quickly.

To address this challenge, the department adopted AES (AI Engineer Staffing) from Akatsuki AI Technologies and brought one senior engineer and two junior engineers, for a total of three, into its projects.

As a result, projects that had previously been difficult to move forward began to progress, leading to the release of an end-user-facing AI feature, the continuation of an R&D project, and the transition to a long-term annual contract. We spoke with Atsunori Nanba, General Manager of the DX Design & Consulting Department, and Mai Kimura, who is responsible for resource procurement, about the decisions they made at the time and how the collaboration worked in practice.

Challenge | We were convinced that AI would become essential, but there was still a shortage of people who could actually implement it

When the DX Design & Consulting Department was first established, interest in AI was high, but there were very few concrete projects. Even so, the department had a clear direction.

Nanba: At the time, there were actually very few AI projects. However, our department’s view was that “AI will definitely become important, so we should actively pursue it.” That was why a company specializing in providing AI engineers caught our attention.

The bottleneck was not simply the number of engineers who “understood AI.” Kimura explained the type of talent the team actually needed.

Kimura: We still had relatively few people, in terms of experience, who could think about how to implement AI into a client’s service rather than simply use AI internally. You need to understand the client’s product and how end users will use it before you can propose the right UI or user experience. Even with something like an AI chatbot, the required level of response accuracy cannot be determined without discussing it directly with the client.

In conventional system development, teams can often agree during requirements definition that “this is the approach we will take.” AI projects are different because the client may also be attempting something for the first time and may not yet know the answer.

Nanba: Clients themselves are often wondering, “Is this really the right approach?” We needed people who could step in and say, “For these reasons, this is probably the best approach right now.” Rather than simply building exactly what they were told, we needed people who could participate in defining the specifications and even discuss business transformation using AI.

As AI projects began to emerge, clients themselves often did not know the correct answer. In that environment, people who could confidently discuss AI-driven business transformation were increasingly important.

The gap that other proposals could not fill

Using external resource partners is common in the system development industry. Even so, the type of talent TechFirm needed was difficult to find in the market at the time.

Kimura: At that time, there were very few companies with a proven track record in AI development, and even when we requested candidates, very few companies could make proposals. There were people with machine learning experience, but almost no one with experience implementing AI into services, or even a strong desire to do so.

Nanba: When you talk about AI talent, you often attract people with a research-oriented mindset. However, the clients we work with require people who can communicate and adapt flexibly. Rather than prioritizing academic interests or the sheer volume of theoretical knowledge, we had already come to believe that younger engineers with sufficient understanding and strong adaptability could sometimes be a better fit.

Selection | Instead of a conventional skills sheet, they came with a proposal

The relationship with Akatsuki AI Technologies began when Akatsuki AI reached out to TechFirm. The deciding factors were the speed of the proposal and the way the candidates were presented.

Kimura: At a very early stage, they showed us information on several engineers and said, “These are the people currently available.” Being able to review the candidates ourselves and determine who might or might not be a good fit was extremely helpful. Normally, once you explain the requirements, it takes time before candidates are actually presented.

Nanba: Instead of simply sending us a standard skills sheet listing technical capabilities, they presented the information in proposal format, which made the key points easy to understand visually. It was also unique that the material included information beyond work history, such as the candidate’s field of study at university.

When presenting AI talent, Akatsuki AI Technologies aims to balance a standardized format that makes candidates easy to compare with enough information to convey each person’s character and background. Because the company originated from an advertising-related business, the readability of its materials has also been designed as part of the value it provides.

When asked whether there were concerns about working with Akatsuki AI Technologies for the first time, Nanba gave a candid answer.

Nanba: Of course there were concerns. However, we had sufficient time for pre-assignment interviews, which gave us a process for resolving those concerns. The project itself was also a new initiative, so whether we had an existing business relationship was not a major factor.

One point both interviewees highlighted as contributing to trust was the importance placed on face-to-face communication.

Collaboration | None of the three engineers waited for instructions

Akatsuki AI Technologies assigned one high-level senior engineer (Engineer A) and two younger high-potential engineers (Engineers B and C), creating a two-tier structure with different roles.

The first noticeable change was how quickly the engineers became part of the team.

Nanba: Everyone began behaving like full members of the project team very quickly. There was no sense of distance, and they integrated into the team fast. They were operating at a level where we could even imagine them taking on a project leader role.

Kimura: Even the younger engineers proactively said they wanted to take on requirements definition, and they continued to deepen their involvement in the project on their own initiative.

Engineer A: Handling a complex R&D project

Engineer A joined a project with a strong R&D component for the client. Rather than having a clearly defined product to build, the role involved working together with the client to determine the direction of the project.

Nanba: Even though the project had a strong R&D element, the engineer handled the difficult coordination very well. We had no hesitation at all about asking them to interact directly with the client.

Engineer B: Version planning and applying AI to the development process

Engineer B joined a software development project for a client’s back-office operations, starting from the phase of building an AI assistant and FAQ chatbot. Over time, the engineer’s responsibilities expanded further upstream.

Kimura: Toward the end, the engineer was even involved in planning future versions. We worked together through trial and error to determine where and how AI could be applied within the development process. The engineer was also very proactive about researching new models and keeping up with the latest information.

Because Engineer B took responsibility for this project, TechFirm was able to reassign internal engineers to other projects. As a result, two additional projects were able to move forward: the development of an image search system and a new business development project for a media company. The impact therefore went beyond simply filling development capacity and helped TechFirm reorganize resources across the department.

Engineer C: Connecting machine learning expertise to real-world service implementation

Engineer C brought expertise in machine learning and applied that knowledge to implementation within a real service. The engineer worked closely alongside TechFirm’s project direction team.

Kimura: Whenever they came into the office, they were almost always sitting together. If something was unclear, they would confirm it, but rather than simply working from instructions like “Please do this,” the discussion was more like, “This is what the client said. What should we do next?” They worked together to think through the next step.

This way of working was particularly effective because there was no single established answer.

Nanba: Even when building an AI agent, there is no established theory or single correct answer. You gather information, test different approaches, and if something does not work, you try something else. I think the engineer was very well suited to that way of working.

A memorable moment

There were also situations in which an Akatsuki AI engineer helped move discussions forward when the client was struggling to make a decision.

Nanba: The client was also uncertain. It was a large organization, so reaching consensus was not always easy. In that environment, I remember the engineer consistently saying internally, “Why don’t we do it this way?” and helping move the conversation toward, “Let’s proceed with this approach.” This was especially true when discussing how to conduct the validation.

Results | Projects moved forward, contracts were extended, and AI features were released

The results of using AES appeared differently across each of the three projects.

1. Continuation of an R&D project and contribution to revenue

Nanba: The project handled by Engineer A is an R&D initiative, so there is no clear endpoint. However, under the condition that the project would end if results were not achieved within a certain period, the work was properly recognized. In that sense, the engineer has also contributed to our revenue.

2. Release of the client’s first end-user-facing AI feature

Kimura: The project Engineer C worked on involved releasing AI to end users for the first time within the client’s service. Being able to deliver work significant enough for the client to issue a press release, while meeting the schedule and quality they required, was extremely valuable.

3. Transition to a long-term contract

In another project, the client came to recognize the need to continuously secure AI engineering resources.

Kimura: For the project involving Engineer B, the client came to understand that continuous development would be difficult unless an AI engineer was secured for a certain period of time. As a result, the engagement was converted into a longer-term contract. The engineer contributed to that outcome as well.

Changes beyond the numbers: Momentum created within a team of peers

In addition to measurable outcomes, both interviewees also mentioned changes in the atmosphere of the team.

Kimura: The younger engineers from Akatsuki AI were close in age to our own engineers, which created a positive atmosphere in which people of the same generation took responsibility and pushed projects forward together.

Nanba: Overall, I felt that many of the engineers had strong communication skills. They fit well with our organizational culture, so we were able to watch the projects progress with confidence.

The DX Design & Consulting Department places importance on creating a positive team environment and generally asks the people it hires to work from the office at least twice a week. During the early stages of a project, an even higher frequency may sometimes be required.

Nanba: Their willingness to respond positively to those expectations was another point we appreciated.

Looking Ahead | Creating the right answer for AI-driven business transformation together

TechFirm has positioned support for client business growth through AI as a company-wide priority. The DX Design & Consulting Department will also continue focusing on creating new value through AI.

Nanba: We continue to have high expectations for support in securing AI talent. Across the industry, business transformation using AI is still at an exploratory stage. I hope we can continue sharing know-how and work together to find the right answers. Physical AI is another area where we would like to develop projects in the future.

The state of AI talent within TechFirm has changed over the past two years. However, Kimura believes an important gap still remains.

Kimura: When we first started working together, there were only a limited number of people internally who understood AI. Training has progressed since then, and more people are now able to use AI. However, ultimately, engineers need to maintain a higher-level perspective: they need to be able to work with clients, use AI effectively, and create something that delivers real value. In that sense, I believe this will continue to be a common challenge not only for us, but for the industry as a whole.

In addition to AI Engineer Staffing (AES), Akatsuki AI Technologies also provides AI-driven business transformation support (AX) and works on projects in the physical AI domain. Given TechFirm’s strengths in XR and devices, exploring new opportunities that combine the strengths of both companies will be one of the next areas of focus.

For companies facing similar challenges

Finally, we asked both interviewees what advice they would give to companies and departments facing challenges related to engineering capacity and resources.

Nanba: What I strongly feel today is that we are moving away from traditional system development. Younger people can now use AI, move upstream into planning and requirements, and still deliver the final product. Rather than relying only on people with conventional career backgrounds, I think it is important to involve younger talent and take on initiatives where there is no established answer. It is very challenging, but also very exciting.

Kimura: From a different perspective than simply bringing in younger ideas, I also think it is important to broaden the range of nationalities involved in order to gain new perspectives. This time, we considered overseas candidates as well before ultimately choosing Japanese engineers, but I do believe there is value in bringing different knowledge and processes into a project.

What kind of company is Akatsuki AI Technologies?

Nanba: They have strong AI expertise while also moving with the speed you would expect from a startup. They are the kind of company you naturally want to support.

Kimura: Their recruiting capabilities are strong, so we can consult with them about a wide variety of AI engineers. Their engineers come from diverse backgrounds, including backend development, and depending on the characteristics of a project, they often have multiple strengths that go beyond AI. Their skills sheets are also detailed, so when we evaluate candidates, it is easy to take an approach such as interviewing two people before making a final decision.