English
JapaneseIMPACT STORIES
Training in Foreign Languages (TIFL)
Training enables consistent, confident global performance
Company Name
FUSO
Industry
Automotive
Location
Japan
Impact
AI-driven global workforce enablement
Application
Web Application
Background
Mitsubishi Fuso Truck and Bus Corporation (MFTBC) operates in a highly regulated and technically intensive automotive environment, where employee training and certification readiness play a critical role in maintaining quality, compliance, and operational excellence. A significant portion of MFTBC’s workforce participates in exam-oriented training programs governed by standardized content and evaluation formats. These programs are essential for skill validation and readiness but are traditionally delivered in a single primary language, creating challenges in a globally distributed and linguistically diverse workforce accurate, while training material was comprehensive and technically language proficiency increasingly influenced training outcomes, leading to uneven participation, slower comprehension, and higher manual effort for training teams responsible for content assessment management. MFTBC required a solution that could improve learning effectiveness without modifying standardized content, while also ensuring scalability, governance, and operational reliability suitable for an enterprise environment.
Challenges Faced
Business Challenges
Language Barrier: Training outcomes were impacted by language proficiency rather than subject knowledge
Manual Content Effort: High manual effort required to create and maintain multilingual training material
Static Assessment: Static mock exams failed to adapt to individual learning progress
Limited Scalability preparation and assessment: Limited scalability of training programs across regions
Low Engagement: Inconsistent learner engagement and completion rates
Technical Challenges:
Mixed Content Formats: Source training material existed in structured and unstructured document formats
Manual Question Creation: Question creation and translation required
significant manual effort
Non-Adaptive Assessment: Lack of adaptive difficulty resulted in
repetitive assessments
Multilingual Accuracy: Ensuring semantic accuracy across
languages, particularly for Japanese
content
Explainable Learning System: Requirement for an explainable system suitable for exam-oriented learning
Solution
Cubastion designed and implemented TIFL (Training in Foreign Languages), a GenAI-powered internal training and mock examination platform that transforms static training material into an adaptive, multilingual learning experience. The platform was built to operate entirely within MFTBC’s enterprise ecosystem, ensuring security, scalability, and alignment with internal IT standards.
Document-Driven Training Intelligence
Training documents are uploaded once into the system
AI processes documents to extract domain specific knowledge
Automatically generates: Questions,
Answer options, Detailed explanations
This eliminated the need for manual question creation and content duplication.
Adaptive Assessment Engine
Uses an FSRS-based algorithm to dynamically adjust question difficulty
Prevents repetition of questions
Focuses on reinforcing weak areas
Maintains exam-oriented assessment discipline
Multilingual Learning Experience
Questions, answers, and explanations are translated dynamically
Learners can select their preferred language
Source content remains unchanged and authoritative
Explainability-First Design
Every question includes:
Explanation of why the correct option is
correct and explanation of why incorrect options are incorrect
Timer pauses during explanation to reinforce learning over speed
Business Outcome
Reduced Manual Effort
~80% reduction in manual effort for content preparation
Faster Test Creation
~60 faster test creation and refresh cycles
Quicker Evaluations
~70 reduction in evaluation turnaround time
Higher Completion
~2x improvement in learner participation and completion rates
