To build an AI learning plan for a department, start from the work the team actually does, not from the tools. List the recurring tasks, decide which roles need which level of skill, teach in a sequence that ends in real practice on those tasks, and agree a few simple measures up front. A plan like this fits on one page and can be funded in part through HRD Corp if your company is eligible.
Start with tasks, not tools
Ask the department head and two or three team members to list the tasks that repeat every week or month: drafting emails and reports, summarising meetings, cleaning data, answering common questions, preparing proposals. Mark which tasks are slow, which are error-prone and which are sensitive because they touch customer or staff data. Those marks decide where AI is worth learning first and where you need rules before anyone experiments.
Define skills by role, not by department
A department is rarely one audience. A useful split is three levels:
- Everyday users. Most of the team. They need to write clear prompts, check outputs and know what must never be pasted into an AI tool. See how to write a good prompt for Claude.
- Champions. One or two people per team who go deeper, build reusable prompts or projects, and help colleagues. They become the first point of contact after training.
- Leads and managers. They need to judge where AI fits, review its output with discipline and set expectations, without needing to build anything.
Technical staff may need a further track, such as workflow integration or certification preparation. Our Claude Certified Associate workshop is one option for people who want a recognised credential issued by Anthropic.
Sequence the learning
A sequence that tends to work is short and cumulative:
- Ground rules first. Cover safe use, confidentiality and your company's usage policy before anyone starts. Our guide to what a company AI usage policy should include helps here.
- Core skills. Prompting, giving context, asking for structured output and verifying answers.
- Task practice. Each person applies the skills to two or three of the department's own tasks, using real (but permitted) material.
- Reuse. Champions turn what worked into shared templates the whole team can use.
- Review. After a few weeks, look at what stuck, what did not and what to teach next.
Keep the early steps close together. Skills fade quickly when people leave a session and wait a month to try anything.
Make practice the main event
Most of the time should be hands-on. Reading about AI changes little; using it on your own reports, emails and spreadsheets changes habits. Give each person a small, real assignment between sessions and ask them to bring one example of something that worked and one that did not. Failures are useful: they show where checking is essential. Our article on checking whether AI output is accurate gives a simple routine to teach.
Measure progress without inventing numbers
You do not need a dashboard. Choose a few measures that the department already understands, and record a rough baseline before training starts:
- Time taken for two or three named recurring tasks, before and after.
- How many team members use AI on a typical week, based on a short check-in rather than surveillance.
- Quality: do managers need fewer corrections on AI-assisted drafts?
- Incidents: any case where confidential data was handled wrongly. This should stay at zero, and each case should lead to a rule or a lesson.
For a fuller approach, read how to measure AI training ROI. If the department also needs to prove value before a wider rollout, a small AI pilot is a sensible bridge.
Plan for HRD Corp funding early
If your company is a registered HRD Corp employer, part of the cost of eligible training may be claimable from your levy. Claimability is subject to eligibility and HRD Corp approval, and approval often has to be in place before the training date, so raise it at the planning stage rather than after booking. Our HRD Corp claim guide covers the usual steps. Agmo Academy is an HRD Corp registered training provider and can run in-house training quoted per scope, tailored to a department's tasks. Prices are subject to SST.
A one-page plan you can reuse
Write down: the department's top tasks, who sits at which skill level, the learning sequence with dates, the champions, the three measures and the review date. If you can fill that page in an hour, you have a plan. If you cannot, the gap shows what to clarify with the department head first. For other planning formats, see our 90-day AI plan and the overview of AI training in Malaysia.