Sam Wong.黃力桐Let’s talk training

TRAINING DECISIONS

AI workshops for Hong Kong teams: talk, half-day, full-day or a programme over several weeks? (2026)

Published: · Last updated:

Short answer: pick the format by what you need people to do afterwards. A 30- to 45-minute talk changes how a room thinks about AI. A hands-on module of about three hours gets people using AI on one real task. A full-day or two-day workshop builds one complete piece of work. A series of sessions over several weeks, department by department, is the format that changes daily work. All of them can run on-site in Hong Kong or online, in Cantonese or English.

By Sam Wong (黃力桐), a corporate AI trainer in Hong Kong who has trained 11,000+ people across 70+ organisations in 13 countries. Every example below is one of my own engagements; the case pages carry the details.

The formats compared

Format Length Best for What people leave with Example from my work
Talk or panel 30 to 45 minutes A large audience, a staff day, a lunch-and-learn, an industry event A shared starting point, the data-safety rules, one workflow to try Arup University: one 45-minute lunch-and-learn for the Hong Kong team. HKCT: about 400 staff at each Staff Development Day
Hands-on module About three hours One team, one tool, one kind of task One real task done with AI, checked and ready to reuse CLP Power: an afternoon module building AI agents inside the company's own Copilot. Chow Tai Fook: market research to a pitched design proposal in one afternoon
Full-day or two-day workshop One or two days A team that needs to finish a complete piece of work A finished output, such as a proposal, a set of memos or a pitch The Hong Kong Jockey Club: two days for management trainees. Chow Tai Fook's product development teams: a full-day creative AI workshop in September 2025
Short evening series Two evenings Professionals who can only train after work Evening one shows; evening two is hands-on The College of Surgeons of Hong Kong: two evenings, 86 valid enrolments
Online series Several sessions Staff spread across offices or countries The same skills, at scale A major Hong Kong bank: five online sessions, 1,500+ participants (attendances), 13 countries
Programme over several weeks Many sessions, department by department Changing daily work across a company Each department's own workflows, built on its own tasks Garden: 20 sessions, 7 cohorts, from HR to R&D. HKCT: a monthly AI learning community under a 24-month agreement

How to choose

Three questions settle most decisions:

  1. What should people do on Monday? If the answer is "think differently about AI", a talk is enough. If it is "use AI on this task", book hands-on time.
  2. How many people, and where? A room of 400 needs a talk. Hands-on practice works best in a small group; I explain why in why my best AI workshops have thirteen people. Staff in several countries usually means online.
  3. Is this the start or the whole plan? One session rarely changes habits on its own. The programmes that changed daily work ran over weeks, one department at a time. I wrote about the reasons in why AI training doesn't stick.

What every format has in common

FAQ

How long should a corporate AI workshop be?

It depends on the outcome you need. A 30- to 45-minute talk sets the starting point; about three hours is enough for one real task; one or two days for a complete piece of work; a series over several weeks to change daily work.

Can AI workshops run in Cantonese?

Yes. I teach in Cantonese or English, with Traditional Chinese or English materials.

Can the workshop run online?

Yes. Every format can run on-site in Hong Kong or online. The bank programme above ran as five online sessions for colleagues in 13 countries.

How much does an AI workshop cost?

Prices depend on length, group size and how much is adapted to your team. I compared providers' published prices in how much corporate AI training costs in Hong Kong.

Next step

Tell me what your team should be able to do afterwards, and I will suggest a format: corporate AI training in Hong Kong.