Case Studies
Check out our project demos and materials. We just ask for a quick form before downloading.
Autonomous AI Agent — Overview
Our autonomous AI agent service specializes in IT operations & maintenance automation. The agent automatically accesses relevant system data in response to inquiries or work instructions, summarizes incident details, proposes solutions, and executes tasks upon human approval.
BPO Auto-Entry Agent (Rakuraku Seisan)
AI agent demo: automated expense entry into Rakuraku Seisan system
Copilot Studio Business AI Solution
Unified automation of 7 BPO processes (My Number registration, savings plans, attendance, payroll inquiry, travel expenses, commuting costs, payroll calculation) on Teams via Microsoft Copilot Studio × Power Automate. Combines Document Intelligence OCR with AI Agent reasoning for end-to-end processing from application intake to back-end system registration.
Aegis AI — Zero-Trust AI Gateway
Security gateway preventing sensitive data leakage in enterprise AI usage. Detects 19 types of Japanese PII (My Number check-digit verification), model choice (GPT-4o/Claude/Gemini), all-inclusive ¥1,500/user/month. No E5 needed.
Vex — AI Document Review Solution | Multi-Cloud & Multi-Model
Vex (Verify + Extract) is a document review solution combining generative AI with rule-based checks. PoC built on AWS (Amazon Bedrock / Claude Opus & Sonnet) achieved 97.4% accuracy and 72% time reduction. Architecture is cloud-agnostic — deployable on Azure, GCP, and supports GPT, Gemini, and other LLM providers.
Generative-AI PoC for Paper Search, Evaluation & Policy Alignment
A PoC using generative AI to streamline academic-paper web search and screening, consistency checks against a knowledge base, and extraction of contextually similar papers. Environment designed and built on Google Cloud; the three steps (search → evaluate → policy alignment) are automated with RAG + LLM.
Harness Engineering (Autonomous AI Control Framework)
Our in-house core edge that hit 97.4% accuracy and 75% token reduction on Vex. Not "build a better model" but "design an environment where AI runs safely, accurately, and cheaply."