Case Study

Client Case Study

Integrating AI Speaking Features
into Existing English Learning Materials

Light Education
×
susuROBO
=
Aleva!

Overview

Operator

Kyoiku Kaihatsu Publishing

Partner: Light Education Inc.

Service

Aleva!

Powered by Clouti

Deployment

AI Tutor: Speaking drills, pronunciation scoring, real-time feedback

Development Period

Approx. 6 months

Environment

In Production

About Aleva!

Aleva! is an online English conversation service designed for beginners. It is operated by Kyoiku Kaihatsu Publishing, a company with over 50 years of experience in educational material development, with Light Education Inc. participating as a partner.

Level × Scenario Materials

5 TOEIC levels × 7 business scenarios

10- or 25-Minute Lessons

Quick sessions that fit any schedule

Avatar Mode

Practice without showing your face

100 Grammar Videos + Shorts

Self-study content linked to lessons

susuROBO's AI dialogue platform "AIROID®" was integrated into Aleva!'s service as an "AI Tutor" feature, automating and enhancing speaking practice.

Leveraging the existing curriculum database structured by 5 TOEIC levels and 7 business scenarios, we embedded AI-powered real-time pronunciation scoring and feedback, creating an interactive and scalable learning experience that works in concert with live tutor lessons and short video content.

Aleva! AI Tutor speaking practice screen

Aleva! AI Tutor speaking practice screen

Challenge

Aleva! is a structured English learning platform featuring materials organized by 5 TOEIC levels × 7 business scenarios, 100 grammar lessons, short videos, and a team of Filipino tutors (95% full-time employees). However, the following challenges existed.

Existing Challenges

  • Speaking practice was heavily dependent on live tutors
  • Difficult to ensure consistency in pronunciation assessment
  • Immediate feedback was hard to deliver
  • Scaling speaking instruction required hiring more staff
  • Rising operational costs

Requirements

  • Real-time pronunciation assessment
  • Immediate, natural feedback
  • Architecture capable of thousands of concurrent users
  • Seamless integration with the existing curriculum database

Solution

susuROBO deployed AIROID® as a "Speaking Drill Engine," building it as a separate AI dialogue layer decoupled from the existing content infrastructure.

Separation-of-Concerns Architecture

Light Education
susuROBO
Level × Scenario Material DB 100 Grammar + Short Videos Tutor & Learner Mgmt Aleva! Frontend Material Database Conversational AI & Avatar Pronunciation Scoring Engine Feedback Generation Session Management API API

Minimal impact on
existing systems

Rapid integration
via API

Full preservation of
curriculum assets

Ensured future
extensibility

Implementation

Real-Time Pronunciation Scoring

  • Speech analysis via ASR (Automatic Speech Recognition)
  • Sentence-level accuracy evaluation
  • Instant feedback display
  • No-speech and silence detection

AI Encouragement & Coaching Feedback

  • Dynamic feedback generation based on context
  • Natural, conversational tone
  • Interactive practice with voice-driven avatars
  • Reduced psychological burden on learners

Session Management API

  • Session-based architecture
  • Usage time and activity tracking
  • Separate Dev / Prod environments
  • Mobile-ready UI integration

Mobile-Optimized UI

  • Responsive design
  • Simple, intuitive user flow
  • Suitable for both children and adults

Results

Since going live, the following outcomes have been achieved.

Speaking practice at scale for thousands of users

Reduced dependency on live tutors

Improved learner retention through instant feedback

Visibility into usage data

Streamlined operations for speaking instruction

Aleva! has evolved into a next-generation English learning platform with AI speaking drills integrated into its existing curriculum.

Monitoring & Analytics

Custom dashboards are built for performance analytics — usage trends, session counts, latency metrics, and more — giving the team real-time visibility into the service in production.

AIROID usage and performance analytics dashboard
Custom analytics dashboard — click to enlarge

Significance

Many educational services already have:

  • High-quality learning materials
  • Well-structured curricula

Yet they often lack:

  • × Automated speaking practice
  • × Conversion of existing materials into voice-avatar-driven interactive drills
  • × Architecture for large-scale concurrent usage

This case study demonstrates that AI speaking capabilities can be retrofitted while preserving existing curriculum assets.

Technical Highlights

Session-based API design Cloud-scalable ASR processing Modular feedback generation engine Dev / Prod environment separation Production-grade infrastructure

Development Timeline

Phase Duration Details
Requirements & Design ~1.5 months API design, UX design
Implementation & Integration ~3 months Conversational AI avatar integration, feedback implementation
Production Readiness & Optimization ~1.5 months QA, load testing
Total ~6 months Design through production launch

About AIROID

AIROID is a modular AI dialogue platform developed by susuROBO. It is designed to integrate existing service content into interactive, voice-avatar-driven drills and exercises.

Supported Domains

Education Elderly Care Retail & Hospitality Entertainment Enterprise Dialogue Systems

Inquiries about this case study

hello@susurobo.jp