The situation
AI was already being tried, but the organization could not manage or repeat the work
In early 2026, Kang Hsuan Educational Publishing Group and Kuse began an organization-wide AI rollout covering the group's education-publishing units and three schools. More than 600 employees and teachers were included.
The question was no longer whether to adopt AI. Staff had already tried several tools, and some people were using AI on their own. The real challenge was making that work dependable, secure, reusable, and manageable across the organization.
Connect proprietary content safely
Curriculum materials had strict access and privacy requirements. Teachers needed AI outputs tied to a specific textbook, unit, and teaching schedule.
Turn individual output into reusable work
Useful prompts and results stayed on personal computers, so colleagues often rebuilt the same workflows from scratch.
Give every role hands-on guidance
Teachers and administrative staff needed concrete methods for their own tasks, not generic AI training.
Manage multiple units as one organization
Each publishing unit and school needed separate accounts, permissions, budgets, and data, with central oversight at group level.
Impact TL;DR
- One company account coordinated four independently managed channels.
- Proprietary curriculum content became a controlled source for lesson plans, handouts, and exam questions.
- Role-specific workshops moved adoption from account setup to daily habits.
- Average time spent on repetitive work fell by 30-40%.
- The rollout became a repeatable model for adding more departments and schools.
01
Rollout model
Central control, local independence
The problem: A single shared workspace would have blurred permissions, costs, and data access across very different units. Four completely separate deployments would have made group-level oversight difficult.
How the collaboration worked: Kang Hsuan and Kuse set up one company-wide account with four independent channels: one headquarters channel for administrative, planning, and editorial staff, plus one channel for each of three schools. The company sets shared policies once, giving central teams visibility and control, while every channel manages its own budget, permissions, data, and recurring AI workflows.
Company account1
Shared policies, account controls, and oversight
Headquarters1
Reports, presentations, planning, and editorial work
Schools3
Lesson preparation, exam questions, handouts, and student support
When another school or department joins, Kang Hsuan can add a channel inside the existing company setup. The management model and onboarding process carry over without starting a new system project.
02
Core capability
Connecting curriculum content to control AI outputs
The problem: General-purpose AI could produce plausible teaching material, but it could not reliably match a particular textbook edition, chapter, difficulty level, or teaching schedule.
How the collaboration worked: Kuse connected Kang Hsuan's proprietary curriculum library directly to the workspace. Teachers select the relevant chapters before creating lesson plans, handouts, or exam questions. The AI works from the chosen textbook, unit, and difficulty level, keeping outputs closer to the publisher's teaching materials, terminology, internal standards, and intended teaching scope.
This connection moved AI from a personal experiment into a working part of lesson preparation. It also established a standard way for Kuse to connect controlled external content sources for future enterprise customers.
03
Platform foundations
Four layers made the rollout manageable
Curriculum connection
Approved teaching materials guide lesson plans, handouts, and exam questions.
Company accounts and permissions
Company, team, and individual account levels let roughly 600 users inherit the right access without account-by-account setup.
Enterprise service limits
Project terms became working settings for model access, paid features, publishing, and file uploads.
Cross-workspace collaboration
Teachers can use the files, conversations, workflows, and shared content they own across personal and team workspaces.
The collaboration layer required more than ten related product changes in one month, covering files, conversation history, published pages, shared content, automated workflows, assistant access, and session continuity when switching workspaces.
04
Adoption
A 90-day path from setup to everyday habit
- M1
System setup
Set up four channels, accounts, permissions, and user guides. Staff and teachers could join their team in three steps.
- M2
Training and habits
Run workshops around real tasks for staff and teachers, helping each role build useful methods and repeatable habits.
- M3
Proving the impact
Track users, active rate, hours saved, and output produced, then turn the strongest examples into a rollout playbook.
Role-specific workshops were central to adoption. Training was part of the service plan and part of what the rollout measured, rather than a one-time handoff after account creation.
05
Daily use cases
Staff and teachers built AI into recurring work
For staff
- Interactive data reportsTurn static tables into clearer web reports for clients and managers.
- Editing and proofreadingPolish copy, adjust tone, rewrite summaries, and tidy documents.
- Slides and designCreate presentations, images, and proposal assets without a design background.
For teachers
- Targeted questions and handoutsCreate material for a selected textbook, unit, and difficulty level.
- Interactive quizzesTurn classroom exercises into activities that make understanding easier to check.
- Learning-gap supportUse wrong answers and test results to create practice for different levels.
What changed after the rollout
AI became deployable, manageable, and repeatable
- Four channels and more than 600 users went live in 90 days.The program moved from initial setup to regular daily use.
- Repetitive work took 30-40% less time on average.The biggest gains appeared in reports, documents, presentations, and lesson preparation.
- Mature workflows saved 40-60% of the time previously required.Templates and stable steps increased the return as teams reused them.
- Normal daily usage cost less than $9 per user each month at 600-person scale.The rollout cost about $5,000 per month while reducing a substantial share of repetitive work.
- New units can join without a new implementation.Kang Hsuan adds another channel under the existing company account and reuses the same management and onboarding model.
About the customer
Kang Hsuan Educational Publishing Group
Kang Hsuan operates education-publishing businesses and schools in Taiwan. This rollout covered its publishing units and three schools, bringing managed AI access and role-specific onboarding to more than 600 employees and teachers.
Education publishingThree schools600+ staff and teachersFour managed channels