shinoi tech
shinoi tech
shinoi tech
The Future of Business through Deep Tech Insight
企業の未来を技術の深層から
Data Dynamism is an approach that views data as something that changes dynamically. This concept builds upon "Data Dynamics*1," a concept proposed in the 1990s by Shigeki Watanabe*2 — whom our Rep Kozuka's mentor — and has been systematized as our company’s perspective on data.
In the real world, most practical data infrastructures treat data as static. Values recorded in tables are implicitly treated as unambiguous “established facts (Facts),” so those who reference them do not question them and make decisions based on those values.
However, are those values truly established?
First, they change over time. Inventory levels, customer addresses, and order statuses can all take on different values the very next moment.
Second, they change depending on how they are derived. The figure for “sales” is a function of rules—such as what constitutes a sale and how returns and discounts are handled—and if those rules change, different figures will be derived from the same reality.
And third, they change depending on the perspective. The sales figures cited by the accounting department, the sales department, and the executive board are all correct, yet they are all different.
At our company, we treat data dynamism not as a secondary phenomenon, but as a primary focus in the design, operation, and evaluation of our systems.
*1 Article: 'From Database To Datadynamics'
*2 Recent publication: 「日本語構造*3」 (Amazon Kindle Edition)
*3 'Japanese Language Structure'
Data Dynamism
Design Focused on Data Ambiguity
AI-driven development
AI is a mirror and an amplifier
“AI is both a mirror and an amplifier. It amplifies both an organization’s strengths and weaknesses”—this is the conclusion of the 2025 DORA report, “State of AI-assisted Software Development.”
In other words, AI does not solve problems; rather, it amplifies even the weaknesses that already exist within an organization. Therefore, simply implementing AI tools is not enough to reap the significant benefits of high productivity.
Since AI tools generate a massive volume of deliverables, it is essential to have a system in place to verify them at the same pace, move those that meet quality standards to the next stage, and return those with defects to the correction and redevelopment loop.
What is needed is not just tools, but a redesign of the entire development process. First, clarify the decision-making flow—such as “who makes which decisions” and “what constitutes completion.”
This applies not only to routine tasks but also to all activities related to decision-making itself—such as presenting design proposals and analyzing the scope of impact—which must be incorporated into the process.
We will begin by undertaking this process redesign.
Project Management in AI-driven Dev.
In project management to date, the practice has been to regard the consumption of man-hours as progress. However, if AI significantly reduces implementation costs, the correlation between man-hours and progress will be fundamentally disrupted.
The traditional assessment of “80 percent complete” was a concept that encompassed not only implementation but also verification of correctness; however, that meaning has been lost.
From now on, the remaining 20%—namely, “verification of correctness,” “system integration,” and “clarification of accountability”—will dominate the entire project. The question we must ask is not “Is the work finished?” but “To what extent can we say with confidence that this deliverable is correct?”
The volume of unverified output is not progress; it is debt. What is even more serious is that this debt accumulates exponentially if left unaddressed. This is because unverified deliverables pile up in subsequent phases, causing correction costs to increase exponentially.
Our company measures progress not by man-hours but by “verified acceptance criteria,” and we redesign every phase of the process—from requirements definition to operation.
In doing so, we build processes that can detect changes in prerequisites early on and respond flexibly, thereby unlocking the true benefits of AI-driven development.