# Twenty sessions and seven cohorts, taking AI into daily work one department at a time.

> Garden, the Hong Kong food manufacturer, started with HR, then extended the programme to sales and marketing, manufacturing operations, finance, purchasing, R&D and senior management. Each cohort practised on exercises built around its own work, using fictional data only.

- URL: https://hyperfocusam.com/en/cases/garden-ai-pioneer-programme/
- Language: en
- Author: Sam Wong (黃力桐), https://hyperfocusam.com/
- 中文版: https://hyperfocusam.com/cases/garden-ai-pioneer-programme/ (Markdown: https://hyperfocusam.com/cases/garden-ai-pioneer-programme/index.md)

---

Garden · Corporate AI training · Department-by-department programme

Garden, the Hong Kong food manufacturer, started with HR, then extended the programme to sales and marketing, manufacturing operations, finance, purchasing, R&D and senior management. Each cohort practised on exercises built around its own work, using fictional data only.

- **When** First workshop July 2025; department programme January–June 2026
- **Audience** 100+ colleagues across about 13 functions
- **Format** 20 in-person sessions across 7 cohorts in 2026

## The brief

The company's IT directive allowed only Microsoft Copilot on the web and Perplexity, and classified data into three tiers, with the most confidential tier barred from AI tools altogether. Some colleagues had paid Copilot licences and some did not, so versions differed. The programme had to work inside those limits.

## How it was designed

1. Each department formed its own cohort. Exercises followed that department's real working situations, but every data set was fictional. Sessions opened with the three-tier data-security card, adapted to each department.
2. Content included three Copilot settings (Think Deeper, custom instructions and memory), the ROCKS prompting structure, a Plus/Minus/Multiply/Divide Copilot framework for finance, and an end-to-end design-thinking exercise for R&D.

## What came out

- **20** sessions
- **7** cohorts
- **4.54/5** trainer clarity (first cohort rating)
- After the first cohort, the company commissioned three further rounds of training in 2026, then asked for a proposal for the next phase.
- Seven of the 19 colleagues in the first cohort volunteered to champion AI in their own departments.

## Feedback

> I can photograph documents straight from my phone, with no special trip to the photocopier to scan them.

— Participant, Garden · post-session survey · translated from the original · January 2026

> You can set up a dedicated agent and stop re-typing the same input every time.

— Participant, Garden · post-session survey · translated from the original · January 2026

> Set up an agent, and effective use of prompt to fine tune the roles of the agent for more concise results

— Participant, Garden · post-session survey · January 2026

## What followed

- In August 2026 I sent HR a recap report covering six months of training by department, built from session recordings, in-class questions and both surveys, as the basis for the next phase.

Next step

## Planning something similar for your team?

Tell me about your team’s work and where it gets stuck. We can decide the right format from there.

[Discuss your training needs](https://hyperfocusam.com/en/#enquire)

[More on corporate AI training](https://hyperfocusam.com/corporate-ai-training-hong-kong/)

[Corporate AI training in Hong Kong: five questions to ask a provider](https://hyperfocusam.com/blog/five-questions-corporate-ai-training-hong-kong/)