EDB AI Education Grant (『智』啟學教撥款計劃) · Use by 31 August 2028繁體中文

School-based teaching desk · Hong Kong primary & secondary

One school day,
one data layer underneath

Planning, papers, marking, guidance, notices and year-group trends on one timeline. The data layer belongs to the school.

School owns the data · Bulk export any time · Free to switch providers · Your own domain
Thursday, an S3 maths teacher
One account, one set of curriculum data, entered once.
  1. 07:50Morning briefOvernight risk list — three students need following up
  2. 08:40PlanningQuadratics: plan, tiered worksheet, a set of questions to ask
  3. 10:15Live modeThree items projected; the class answers on phones with a confidence rating
  4. 11:00Guided practicehints ≤ 2Two students stuck after class; hints capped at level two
  5. 14:20MarkingLast week's quiz photographed; AI drafts per-item scores, the teacher confirms
  6. 16:30Panel meetingYear-group means and mastery trend, projected as-is — no separate report

Thursday in one line: plan → guide → mark → notice

At a glance
15
functional areas in production
800 + 100
student and staff accounts, not charged per seat
0
additional token charges
4
AI gates (input / system / output / audit)
  • Interface: Traditional Chinese (HK) / English
  • Primary and secondary configured separately
  • The school owns the data and can export all of it
  • Runs on a school server; works on your own domain
01On the ground

Four things a general chatbot cannot do

  1. 01

    It does not know your lesson

    To set an S3 quadratics worksheet, you first need to know which textbook, which sub-section, and how this class did last term. That lives in the school, not in the model.

  2. 02

    It has no reason to hold back

    A student pastes the question and has a full solution in three seconds; what was learned is how to paste. Holding back has to be enforced in the system, not left to a teenager's self-discipline.

  3. 03

    The record sits in someone's personal account

    Prompts, marking habits and school materials accumulated in a personal account walk out of the door with the teacher. School knowledge only compounds if it is written to the school's database.

  4. 04

    Nobody can see what happened

    Who used how much, what students asked, which reply a gate stopped — without an audit trail a principal cannot answer the IMC, or write the grant report.

02System

Fifteen areas, one line of data

03Data

Four gates, each one logged

  1. 01

    Inbound

    What the student typed is read first — inappropriate requests, personal data and jailbreak attempts stop here, with a lower threshold in student mode.

  2. 02

    System instruction

    School policy, age band, subject context and the no-answers rule go into every call, rather than being left as a reminder to the student.

  3. 03

    Outbound

    The reply is classified again; anything questionable is masked and queued for an administrator to review.

  4. 04

    Audit

    Every call records module, model, usage and gate action; quotas are set per role and per user.

The line-by-line list
04Four clauses

The data belongs to the school, and it is in the contract

These four are contract clauses, not marketing lines, and they apply the same on-premises and in the cloud. The exit has to be wide before a school can reasonably put its own knowledge in.

  1. 01Ownership

    Plans, resources, the knowledge base, assessments, student records and school analytics — intellectual property and ownership rest with the school. Edor.ai acts as data processor on the school's instructions.

  2. 02Download

    One click in the admin panel exports everything: one JSON per table, resources and the knowledge base as Markdown, original files attached as they are. Free, unlimited, no request needed.

  3. 03Leaving

    A school may take its data to another provider at any time without giving a reason. At contract end we assist the migration and provide a deletion certificate.

  4. 04Front door

    Reach the system on the school's own domain (e.g. ai.yourschool.edu.hk) with a single CNAME record; the login page carries the school crest and name.

On a school server, or in the cloud ai.yourschool.edu.hk CNAME edor.ai
05Onboarding

Four weeks from nothing to whole-school

  1. 01

    Week 1 · Move the data

    School, year, classes and accounts by CSV; curriculum units and objectives loaded; AI provider chosen and the safety policy written.

  2. 02

    Week 2 · Teachers first

    On-site workshop: planning, setting and marking, how to design a guided task. The panel co-plans and produces the first school resources on the spot.

  3. 03

    Week 3 · Students in

    Open one or two classes; teachers issue the first guided task and assignment, and adjust the hint ceiling while watching the transcripts.

  4. 04

    Week 4 onwards · Read the numbers

    Run the PIE review on real platform numbers; consultants return to co-plan and observe, and the grant evidence accumulates alongside.

EDB AI Education Grant (『智』啟學教撥款計劃)

Turning a one-off grant into two years of classroom habit

The hard part is not the application — it is having something to show two years later. Edor.ai connects platform, training, on-site consultancy and evidence into one line: teachers teach as usual, and the numbers the report needs grow underneath.

  • Platform and consultancy fall inside the grant's eligible uses
  • Coverage: at least three subjects, two grades each
  • Demo lessons, sharing sessions and training hours logged automatically
  • PIE outputs the interim and final report data directly
Request a grant proposal pack

Grant ceiling

HK$500,000

Demo lessons67%
Sharing33%
3 subjects × 280%
Grant amount
Up to HK$500,000 per school
Launched
16 December 2025
Use by
31 August 2028
Funded by
Quality Education Fund
08Questions

The fourteen questions schools ask most

Edor.ai is a school-based teaching desk for Hong Kong primary and secondary schools. Planning, setting and marking, guided practice, the resource and school knowledge base, the student study area, the PIE cycle and SDAS results analysis all share one data layer — and that layer belongs to the school. Training and on-site consultancy come with it, aligned to the EDB AI Education Grant.

No. Guidance runs in four levels: clarify the question, point in a direction, demonstrate part of it, and only at the last level — full steps — if the teacher has opened it. Teachers set the ceiling per task, and can read transcripts live, pause a student or flag a message.

Any subject works, and unit templates ship for Chinese, English, Mathematics, General Studies / Science, Humanities and STEM. The interface is Traditional Chinese (Hong Kong) with an English toggle; AI output language is set per tool, and oral practice covers Cantonese, Putonghua and English.

Each school gets its own database and file volume, which can sit in Hong Kong cloud or an environment the school nominates. Administrators can export the knowledge base, resources, curriculum, rubrics and student records in bulk at any time; at contract end we assist migration and provide a deletion certificate.

The data belongs to the school, and it makes no difference which way you deploy. Either way an administrator can download everything from the admin panel in one click (JSON / CSV / original files), free and without limits. We keep the exit that wide because we would rather keep schools by being useful than by being hard to leave.

See it once, then decide whether it goes in the proposal

Thirty minutes, run live on your school's subjects and grades.