Govern the machine.
Before it governs the file.
Eight modules. Thirty-two lessons. One marked capstone. For the Hong Kong professionals who will be asked to sign off on AI, and who would rather know what they are signing.
AI does not fail loudly.
It fails in the paragraph nobody reviewed.
Most organisations are not reckless with AI. They are busy. A vendor demo, a pilot, a press release. Then a complaint arrives, the privacy commissioner writes, and nobody can say who approved the model, on what data, against which risk.
You use GenAI daily, but cannot say which of your use cases would be "high-risk".
You have read about the EU AI Act. You have not read your own department's AI policy, because there isn't one.
You know "bias" is a risk. You could not name the three places it enters a model.
Your vendor says the model is "secure". You do not know what prompt injection is.
An interview panel asks for your view on AI regulation in Hong Kong. You have opinions. You do not have a framework.
None of these are character flaws. They are what happens when a technology arrives faster than the governance around it. The AI Steward Bundle exists to close that gap — with frameworks, with supervision, and with a credential at the end.
Four pillars.
One steward.
What AI is, how it learns, where data flows.
Harm, bias, risk scoring, the register.
Threats, guardrails, red teaming.
Hong Kong policy, the memo, the panel question.
"Frameworks are free. Judgement is not. The bundle exists to put your judgement under supervision until it holds."
Eight modules.
Thirty-two lessons. One file to defend.
Every lesson follows the IRB "Rice Grain" format: 1 × 12-min video · 1 × reading brief (1,500 words) · 3 key terms · 1 HK case vignette. Plus a six-question check and an Officer's Note — how a civil servant would frame it.
From bystander to steward,
in eight modules and one file.
- M0 · 根基Module 0 · Foundations: Why AI Needs Governing
- M1 · 定義Module 1 · Defining AI Systems
- M2 · 數據Module 2 · Data & the AI Lifecycle
- M3 · 學習Module 3 · How Machines Learn (for non-engineers)
- M4 · 傷害Module 4 · Harm, Risk & Bias
- M5 · 風險Module 5 · AI Risk Management
- M6 · 安全Module 6 · AI Security
- M7 · 施政Module 7 · Policy, Public Sector & the Interview Room
- ★ · 結業Capstone — The Steward's File
Content is the floor.
Coaching is the product.
Each lesson
The Rice Grain format: 1 × 12-min video · 1 × reading brief (1,500 words) · 3 key terms · 1 HK case vignette · 6-question check · 1 "Officer's Note" — how a civil servant would frame it. A civil servant's reading pace, not an engineer's.
Each clinic
A live 2-hour Steward Clinic on the last Saturday of the month — Hong Kong cases worked in the room. Replay included.
The capstone
The Steward's File: a complete AI Risk Register plus a 2-page governance memo for a Hong Kong scenario. Marked against a 5-criterion rubric within 7 working days.
The 1-1
Two 60-minute Zoom sessions with Mike — one to scope your file, one to defend it. Practitioner and Executive tiers.
The certificate
A verifiable IRB credential with a credential ID, valid for two years. Renewed by the annual Steward Refresh.
Four doors.
One of them is free.
Every tier includes the AI Literacy Test and the Module 0 preview. Prices in HKD.
- AI Literacy Test + archetype report
- Module 0 preview
- Modules 0 to 7, all quizzes
- Final exam
- IRB Foundations Certificate
- 12-month Teachable access
- Everything in Foundations
- 4 live 2-hour Steward Clinics (last Saturday monthly, replay included)
- Marked Capstone — The Steward's File
- 2 × 60-minute 1-1 Zoom with Mike
- IRB Certified AI Steward credential
- Up to 5 seats of Practitioner
- 1 private 3-hour in-house AI governance workshop
- Organisation AI policy template review
- · Monthly plan: 4 calendar months, cancel anytime (access to cohort items stops; the certificate requires the capstone).
- · WELCOMEIRB15 (15% off) applies to the Practitioner monthly plan only.
- · Existing Persistency Bundle Companions: 20% loyalty discount on Practitioner (code STEWARDIRB20).
- · Refunds follow the existing Returns & Refunds Policy.
Two credentials.
One earned with a file you can show.
IRB Foundations Certificate in AI Governance
- —Complete all 32 lessons (Teachable completion tracking)
- —Pass the Final Exam: 40 questions, 60 minutes, pass mark 75% (30/40), 2 attempts, 7-day wait between attempts
IRB Certified AI Steward (IRB-CAS)
- —Foundations requirements, plus
- —Attend or watch 3 of 4 Steward Clinics
- —Capstone "Steward's File" scored 70/100 or higher on the rubric
- —20-min oral defence on Zoom (one panel-style question on your file)
- —Valid for 2 years. Renew by completing the annual "Steward Refresh" — a 2-hour update webinar plus a 10-question quiz, HK$480.
- —CPD: 24 learning hours (Foundations) / 36 learning hours (Steward) for self-declared CPD.
"IRB credentials certify completion of Iron Rice Bowl's programme. They are not accredited by, or affiliated with, any government body, the HKSAR Government, IAPP or ISO. They do not confer a professional licence."
Graduates wanting an external credential can continue to IAPP AIGP; the curriculum maps well to its domains, but we make no pass guarantee.
Built by someone who sat on the panels —
not someone who studied them from outside.
Mike Yeung spent twelve years inside the Hong Kong civil service, rising to Senior Administrative Officer across seven bureaus. He briefed Bureau Secretaries. He fronted LegCo Panels. And — critically — he sat on the recruitment panels that decide whether today's candidates will wear the same title he once did. He drafted the kind of policy papers AI is now rewriting — and built this bundle so the next steward does not have to learn the machinery the hard way.
The honest answers.
The model is already in the building.
The question is who is governing it.