“We get vendor proposals, but can’t judge whether they really fit us. We keep ordering at the quoted price.”
“We were told AI would make things better — what we got is confusion and cost overruns.”
“Legacy systems are so entangled we don’t know where to start. The team is wearing down.”
Where we can help
AI & development teams
Adopting AI-driven development
We help you avoid the most common ending: tools nobody uses. We use Claude Code and Codex across all of our own product development and publish our spec-driven workflow as open-source Agent Skills. We rescued a stalled identity-platform project by switching it to AI-driven development, shipping with half the originally planned resources and schedule.
We reduce the risk of breaking what works through assessment and staged migration plans. Deciding what to rebuild and what to keep takes both engineering judgment and investment judgment.
Emerging technology
Unproven domains (Web3, DID/VC, and more)
We take on domains nobody else will — starting from the research. DID/VC was one of them: we learned it after being asked, went from studying the standards to designing patent-worthy use cases to building the platform, and became a named inventor on a registered patent (JP7857517) for AI-agent authority delegation. Bringing an unknown domain up to decision-and-implementation grade is what we do.
Payments & identity
Payments & identity systems
Fifteen years of payment and O2O systems for Sony Group (consulting), JR-affiliated station retail, Kappa Sushi, Japan Post, and others — including compliance with enterprise security standards and third-party penetration testing. We know how to build systems that must not stop.
Selected work
Rebuilding a next-generation identity platform
1/316789012345678901/38901234567890123Team size: 6 → 2 people
The initial DID/VC platform used proprietary specifications that did not comply with W3C standards. Development had stalled before completion.
Our work
Restarted the project from scratch, researched W3C standards, and reassessed the product concept, architecture, and libraries.
Outcome
Introduced spec-driven development, parallel AI sessions, and adversarial review loops. Development time was halved, from approximately six months to three, and the team reduced to one-third, from six people to two. The rebuilt beta enabled customer proposals.
Rebuilding an applicant tracking system
1.5 months16789012345678901.50123456789012345monthsFrom restarting development to the MVP release
Initial challenge
Conducted technical due diligence on an applicant tracking product after two outsourced development attempts failed to reach completion.
Our work
Reviewed the existing architecture and GitHub development history, reported the findings, and recommended rebuilding with a new architecture. At the client’s request, also took on MVP development.
Outcome
Delivered an MVP in one and a half months from the start of redevelopment, after the previous effort had remained incomplete for more than five months.
Requirements definition through delivery management for payments and O2O systems at major station-retail, apparel, restaurant, and postal enterprises.
Business and investment planning
For Sony Group and others, our consulting covered not just system design but business plans and investment plans for new ventures.
Engineer evaluation in the AI era
When AI writes the code, productivity and roles change. Drawing on published experiments from engineering organizations such as OpenAI and the practice of harness engineering, we advised on how evaluation and team structure should evolve.
System and process inventory, risk analysis, improvement plan
Project engagements
AI development workflow adoption
JPY 800k–2M
Claude Code / Codex and CI/CD, from rollout to adoption
PoC / MVP development
JPY 1.5M–5M
Rapid validation of new business, AI, and Web3 products
Ongoing engagements
Technical advisory
JPY 300k–800k/mo
Design reviews, technical consultation, direction setting
Fractional CTO / architect
From JPY 1.5M/mo
Acting technology lead: team building, vendor management, key decisions
Independent review
Technical due diligence / vendor review
JPY 800k–3M
Third-party review of code, architecture, team, estimates, and debt
Initial requirements discussion and quote: free.
Sample assessment and review report
An example of the report provided with our AI-Driven Development & Modernization Assessment or Technical Due Diligence / Vendor Review. This sample covers development progress, team processes, and an improvement plan. The scope and report contents are tailored to the purpose of each engagement.
Scope and measurement
Agree on the systems and review period, then examine pull requests, reviews, and work reports.
Analysis and interviews
Investigate delays and concentrated responsibilities, and ask about work the records do not capture.
Findings and action plan
Receive documented findings, open questions, and an improvement plan with owners and next steps.
All projects, people, and figures are fictional. Screens and report layouts were recreated for this public sample.
Drawing on experience in software engineering and business management, he helps companies select technologies and build development teams based on their business plans, budgets, and development needs. Support also covers investment planning and vendor management.
Started his engineering career developing ATM systems, led e-commerce package development at Digital Garage, and served as a board director and executive officer at e-Cash Inc. (now Pass Co., Ltd.). The company listed on the Tokyo Stock Exchange Mothers market during his tenure. He founded Samurai Inc. in 2010. Following investment from Sony Payment Services, Samurai was acquired by a Web3 company in 2024, and he continues to serve as CEO. He subsequently founded ZenChain LLC, providing technical advisory and development support for AI-assisted software development, system modernization, and in-house development.
Patents
Inventor of JP7857517 (authority delegation to AI agents) and JP7905143 (identity and authenticity verification using DID/VC)
Columns
@IT (solo) and ITmedia (co-authored)
OSS
Spec-driven development Agent Skills (anyoneanderson/agent-skills)
Will our confidential information be exposed to AI tools?
We operate with training opt-out settings, and we've built custom hooks that prevent AI tools from sending confidential information externally, configured per engagement. This will be published as our AI usage policy, so you can review it against your security standards before contracting.
Is the initial discussion free?
Yes. The initial discussion of your requirements and a quote are free. Technical research, assessments, and specific advisory work begin only after you agree to the scope and fees. You can also request a standalone 60-minute consultation or assessment.
Can you work on-site or five days a week?
On-site residency is not offered in principle. In-person meetings are possible unless the location is remote. Five-day engagements depend on available capacity at the time — please ask during the consultation.
Can you do the development itself?
Up to PoC / MVP scale, we build it ourselves. For larger builds, we can help through in-house team design or vendor selection and management — let's discuss.
Let’s discuss your AI adoption or system modernization.