Test experience and intuition against observation and data. Moving between research and consulting, we build systems for better decisions.
OUR THESIS
For us, studying business scientifically does not mean reducing everything to numbers.
It means observing people, organizations, and markets; asking questions; testing hypotheses; and learning from results. We pursue systems that improve decisions through change, not one-off answers.
CURRENT RESEARCH
We observe not only technology and data, but how people interpret, choose, and change behavior. We study the conditions for systems that keep working in practice while protecting decision quality and autonomy.
We treat AI not simply as automation, but as an intelligent system that supports exploration, judgment, and execution. We study the conditions for sustained, trustworthy use in real work—including roles, authority, evaluation, and visibility.
Data must do more than accumulate: its meaning must align, its changes must be read, and its evidence must inform decisions. We study data integration, metric design, analytics, and forecasting as a foundation for organizational learning.
Attention, trust, choice, motivation, and habit all shift with context. We observe how interaction with AI changes human decisions and behavior, then test the conditions for durable improvement without compromising autonomy.
Data supports AI decisions, and AI activity creates new data. Our central theme is making this cycle measurable, evaluable, and improvable.
We measure more than whether an AI suggestion was accepted: the context in which it was trusted, what was chosen, and whether the behavior persisted. The goal is not persuasion alone, but a design that protects understanding, objection, and autonomy.
APPLIED RESEARCH / PRIORITIES
Outcome-producing use, organizational adoption, and vulnerability through reliance: we test these three issues with real data and observed change as AI becomes a business capability.
We measure value through revenue, gross profit, time, and quality—not adoption counts. We define workflow baselines, usage costs, and validation metrics to identify when ROI is reproducible.
We integrate workflow design, education, evaluation, authority, and knowledge sharing so results do not depend on a few experts. We study the conditions under which AI becomes part of everyday judgment and continuous improvement.
We examine declining judgment, cascading errors, hollowed-out knowledge, and blurred accountability. We design controls that let people verify, challenge, and stop the system.
SCIENTIFIC METHOD
Research and consulting move through the same cycle.
Read the field, the data, and the constraints. Separate what is known from what remains uncertain.
Map the relationships in a complex situation and turn them into testable questions and hypotheses.
Build small, examine quantitative and qualitative evidence, and learn from the difference.
Embed what we learn into work, systems, and the flow of decisions.
CONSULTING
Across management, operations, data, and technology, we define the question, build decision-ready evidence, and create systems that work.
We read the market, customers, operations, and data to define the question that matters now.
Starting from return on investment, we design target workflows, an adoption roadmap, and controls for reliance risk.
We test prototypes with real data, then establish evaluation methods and operating conditions.
Start with a question we can test,not a ready-made answer.
INSIGHTS / RESEARCH NOTES
Research on AI, data, behavioral science, and organizations—along with methods worth testing in practice.
LLMs can shift answers toward a user's stance or pushback, and ratings miss it. Investment reviews should not count AI agreement as independent verification, should re-verify AI objections, and should probe how models update at rollout and on change.
Read article
Design evidence for AI-agent actions around a chain of approval, execution, and state change—not full-content retention. Capture disputed content selectively under protection, then test reconstructability and minimization as separate release gates.
Read article
A role-and-output reading of EU AI Act Article 50, translating human-facing disclosure, machine-readable marking, exceptions, and version evidence into one placement record.
Read article
TOKYO, JAPANRESEARCH & CONSULTINGCONTACT
We work on AI and data research, business challenges, and collaborative projects.
