01
Start from the problem
We ask 'What needs to be solved?' before 'What can AI do?' We aim for production systems from day one, not proof-of-concepts that stop there.
Software development, reimagined with AI.
Production AI systems
and cross-border software.
GENGAKUTAKU INC. helps Japanese companies design, build, and operate the next generation of AI applications, global products, and practical digital workflows.
PoC speedValidated in 2 weeks
Operating modelDesign to operation
Market coverageJapan, China, U.S.
Security posturePrivate and on-prem ready
01 / CapabilitiesAI Engineering
LLMs, RAG, and agents in production systems
02 / CapabilitiesCross-Border Product
Simultaneous delivery across Japan, China, and the U.S.
03 / CapabilitiesDX Consulting
From paper and Excel to SaaS workflows
04 / CapabilitiesData Intelligence
Evals, monitoring, and cost as ops design
Services
What we do
A focused engineering practice for AI systems, production software, and cross-border product work.
Track Record
Proof you can evaluate.
Within confidentiality limits, we show the problems we solved, the technical choices we made, and the measurable outcomes delivered.
PoC speedValidated in 2 weeks
Operating modelDesign to operation
Market coverageJapan, China, U.S.
2025Finance
Internal document RAG system — design and build
Designed and built a RAG system over tens of thousands of internal policies, manuals, and inquiry records, enabling natural-language search and answer generation. Deployed on-premise to meet strict data residency requirements with zero external data transmission.
Outcome: Reduced average inquiry response time by 40%. Also shortened new-hire onboarding period.
2025SaaS
Japan-origin SaaS — Greater China localization
Supported the Greater China expansion of a growing Japanese B2B SaaS product. Redesigned the i18n architecture, implemented Simplified and Traditional Chinese support, optimized CDN delivery for mainland China, and ensured PIPL compliance — all in one engagement.
Outcome: Improved page load time in Greater China to under 3 seconds. Local user count reached target within 3 months post-launch.
2026SMB (Services)
Excel-to-SaaS migration and process redesign
Migrated a 20-person service company from Excel-and-paper operations (accounting, attendance, contracts) to cloud SaaS. Completed the full cycle from As-Is process mapping through To-Be redesign, tool setup, staff training, and adoption support in 3 months.
Outcome: Shortened monthly closing from 5 business days to 2. Fully digitized paper-based contract management.
Operating Method
We do not stop at PoC. We design for operation.
Discovery, architecture, implementation, evaluation, and operations are treated as one continuous system. AI adoption and product development move forward with clear decision points.
02
Start small, move fast
PoC in 2 weeks, MVP in 4 to 8. We stack visible results instead of making grand plans.
03
Own the outcome
Delivery is not the finish line. We take responsibility through operations, maintenance, and improvement. We walk the full lifecycle with our clients.
Global
Three markets, one team.
Chinese AI implementation velocity, U.S. large-scale distributed systems design, and Japanese quality standards — woven into a single engineering organization that supports our clients' overseas expansion and new product development.
News
Tech Signals
We track what's happening at the frontier of AI and software development, kept here as our team's reading list.
Hivetrail（Anthropic 報告書解説）Anthropic's 2026 Agentic Coding Trends Report — 8 shifts reshaping engineering teams
Anthropic's report frames the 2026 shift around SDLC compression from weeks to hours, multi-agent architectures, long-running agents, and security-first design as the structural changes for agentic coding.
Read on site
AnthropicAn update on recent Claude Code quality reports — a model engineering postmortem
Anthropic identified and resolved three separate issues affecting Claude Code: a reasoning-effort default change, a caching bug that caused memory loss, and a system prompt modification that reduced coding performance.
Read on site
Google CloudIntroducing the Gemini Enterprise Agent Platform — Google's full-stack bet on enterprise agents
Google Cloud unveiled a unified platform to build, scale, govern, and optimize AI agents. The release combines Vertex AI capabilities with new features for agent integration, DevOps, orchestration, and security.
Read on site
TechCrunchOpenAI updates its Agents SDK to help enterprises build safer, more capable agents
OpenAI added sandboxing to its Agents SDK so autonomous systems can work with files and approved tools in controlled environments while preserving overall system integrity.
Read on site
SiliconANGLECloudflare expands Agent Cloud with new tools to build and scale AI agents
Cloudflare announced Dynamic Workers for secure code execution, Artifacts for Git-compatible storage, Sandboxes, the Think framework for persistent tasks, and an expanded AI model catalog—pushing agents from prototype to production.
Read on site
TechCrunchOpenAI launches Frontier — an enterprise platform to build and manage AI agents
OpenAI introduced Frontier, an open enterprise platform for organizations to build, deploy, and oversee AI agents alongside human employees. Initial customers include Intuit, State Farm, Thermo Fisher, and Uber.
Read on site
Each article's content and copyright belong to its original publisher. We summarize public titles, dates, and context on-site for reference.
About
About us
GENGAKUTAKU INC. reinvents how software is built with AI, and brings that power to the everyday lives of people.