# AI Automation for Tuition Centres: 3 Do's and 3 Don'ts

Updated: 2026-07-17

The best tuition-centre tasks to automate are repetitive, rule-based and easy to correct. Teaching quality, personal data, money, complaints, refunds, non-standard discounts and formal commitments should retain staff review or manager approval.

Use three tiers:

1. **Automatic execution:** sorting, searching and reminders under fixed rules.
2. **AI preparation with staff review:** changes to student records, amounts or outbound messages.
3. **Human decision:** teaching judgement, financial exceptions, complaints, refunds and formal commitments.

## Approval matrix

| Scenario | AI role | Human decision | Tier |
|---|---|---|---|
| Student data and invoices | Parse voice or text, match records, prepare drafts | Identity, class, amount, discount and payment status | Staff review |
| Daily opening brief | Summarise classes, absences, make-ups and tasks | Conflicts and priorities | Automatic summary |
| WhatsApp leave or make-up | Identify intent, check policy and slots, draft reply | Identity, exceptions and final message | Normal cases automatic; exceptions reviewed |
| Teaching materials | Draft questions and answer formats | Accuracy, difficulty, objectives, sources and copyright | Teacher review |
| Complaints and refunds | Organise facts, timeline and records | Responsibility, response, remedy and refund | Manager decision |
| Discounts | Calculate approved offers | Non-standard or stacked offers | Manager approval |

## Three things to do

### 1. Use voice input to prepare student updates and invoice drafts

Match students using an ID or multiple data points, compare requests with current enrolment, fees and payments, mark uncertain data as `pending_review`, and update official records only after confirmation. See [How Tuition Centres Can Use AI to Organise Student Records](https://oneflash.hk/blog/tuition-centre-ai-organize-student-records-hong-kong).

### 2. Generate a daily opening brief

Summarise classes, tutors, rooms, absences, make-ups, waiting lists, unconfirmed transfers, invoices, payments, materials and exceptions. Every item should link to its source, and conflicts should be flagged instead of guessed.

### 3. Let AI handle the first stages of WhatsApp leave and make-up requests

AI can verify the student and lesson, check policy and capacity, offer approved slots, update records after confirmation and send the final message. Staff should take over when identity is unclear, classes are full, fees differ, or the request involves an exception, refund, complaint, health issue or commitment outside policy. Related reading: [How a WhatsApp AI Chatbot Handles New Customer Enquiries](https://oneflash.hk/blog/whatsapp-ai-chatbot-new-customer-enquiry-hong-kong).

## Three things not to do

### 1. Do not publish AI teaching materials without teacher review

Teachers must verify answers, difficulty, teaching objectives, language, sources, copyright, bias and age suitability. The Education Bureau's [Guide to Using AI in Teaching](https://www.edb.gov.hk/attachment/en/edu-system/primary-secondary/applicable-to-primary-secondary/it-in-edu/DEBP/DEBP_EN.pdf) supports a teacher-led, AI-assisted principle. Copyright guidance is available from the Intellectual Property Department's [Copyright and Education](https://www.ipd.gov.hk/en/copyright/faqs-and-guidance-notes/copyright-and-education/) resources.

### 2. Do not let AI decide complaints and refunds

AI may acknowledge receipt, organise allegations, build a timeline and retrieve records. Responsibility, response, apology, remedy and refund remain decisions for authorised staff.

### 3. Do not let AI invent or stack discounts

AI may calculate standard offers under approved rules. Record eligibility, validity, stacking rules, maximum discount, approving role, final amount, approver and reason. Non-standard commercial decisions remain human.

## Before implementation

Define the trigger, read access, write access, normal rules, handover conditions and audit trail. Restrict student and parent data to business need and retain human checks. See the Privacy Commissioner's [AI data-protection resources](https://www.pcpd.org.hk/english/artificial_intelligence/index.html).

## Ten-question test

Is the rule written? Is the source reliable and current? Will AI stop when data is ambiguous? Is an error easy to reverse? Does the step change official records, make a commitment, affect teaching or personal data, or involve money, complaints, refunds or non-standard discounts? Who receives exceptions, and can the centre trace the suggestion, approval and sent version?

Any step affecting official records, commitments, teaching, personal data or money should have staff review; financial exceptions, complaint positions and formal commitments should require manager approval.

## FAQ

### Does every WhatsApp leave request need staff approval?

No. Normal cases can be automated when identity and lesson are clear, policy is followed and only approved slots are offered. Exceptions go to staff.

### Can AI automatically generate invoices or receipts?

AI can prepare drafts from confirmed data; identity, amounts, discounts and payment status should be checked before issue.

### Can AI send messages directly to parents?

Fixed reminders and normal cases may be automatic. Exceptions, money, complaints, sensitive data and formal commitments should be reviewed.

### Can AI process student data?

Start with a defined purpose, restrict read and write access, avoid unnecessary personal data and retain operation records.

### Must teachers review AI-generated materials?

Yes. They should check answers, difficulty, teaching objectives, language, sources and copyright.

### Which workflow should a centre automate first?

Choose a frequent, rule-based workflow supported by existing data and easy to reverse. Daily briefs and WhatsApp leave or make-up requests are safer starting points than complaints, refunds and discounts.