AI Training for Staff: How to Get Your Team Actually Using AI (Not Ignoring It)

You’ve signed up for an AI tool, shown your team a few impressive examples, and waited for things to change. But the team is still writing every email from scratch, chasing the same overdue invoices and spending too much time on admin.

This is a common problem. A tool on its own does not change how people work. Useful AI training for staff gives your team a safe, practical way to try AI on tasks they already do.

Here’s how to get a small team started, without turning training into another big project.

Start with a real task, not a talk about AI

A general introduction can explain what AI is, but it may not help someone write a better customer follow-up or turn meeting notes into a list of actions.

Ask each team member to name one task that is repetitive, time-consuming and low-risk. Good places to start might include:

• Drafting a polite follow-up email
• Turning a set of job notes into a checklist
• Rewriting a service description in plain English
• Creating a first draft of a job advertisement
• Organising non-confidential meeting notes into action items

Choose one task and test it together. The goal is not to replace anyone’s judgement. It is to find out whether AI can help with a first draft or save a few steps.

Run a short lunch-and-learn

A lunch-and-learn works well because it is short, practical and easy to fit around a busy week. Keep the first session to about an hour and use a tool your team is allowed to use.

Avoid spending the whole time on slides. Show one ordinary task, explain the basic rules, then let staff try it themselves with a fictional or de-identified example.

A useful first exercise is drafting a customer follow-up email. Give the AI a brief, made-up set of notes, then ask it to produce a warm, concise email. Together, check what it got right, what it missed and what a person must add before sending.

Make it clear that AI can produce confident-sounding errors. Staff should verify facts, prices, dates, promises and anything that could affect a customer or employee.

Give staff a simple prompt structure

Many first attempts fail because the request is too vague. “Write an email” gives the tool very little to work with.

Teach staff to include five things:

1. Role: What should the AI help with?
2. Task: What do you want it to produce?
3. Context: What background does it need?
4. Rules: What must it avoid or preserve?
5. Format: How should the answer be laid out?

For example:

Help me draft a customer follow-up email. Use the fictional notes below. Keep it under 150 words, friendly and clear. Do not invent prices, dates or promises. List any missing details I should check before sending. Notes: [add non-confidential notes].

Staff can then adjust the draft, rather than accepting it as finished work. This simple pattern can also be used for a quote cover email, a roster handover or a first draft of a procedure.

Build a shared prompt library

If one person finds a prompt that works, save it somewhere the whole team can find it. A shared prompt library helps staff avoid starting from scratch and makes useful examples easier to repeat.

Organise the library around tasks, not AI jargon. Start with a handful of common jobs, such as “customer emails”, “meeting actions” and “procedure drafts”. For each prompt, include:

• What the prompt is for
• The prompt itself, ready to copy
• What information is safe to use
• What the person must check before using the result
• Who owns the prompt and when it was last reviewed

Keep the library small at first. A few tested prompts are more useful than a long folder no one opens. Remove prompts that produce poor or inconsistent results.

Appoint an AI champion

Choose one person who is curious, practical and trusted by the team. They do not need to be a technical expert.

The AI champion can collect staff questions, test prompts, share examples and flag issues to the manager. Give them time to do this as part of their role, rather than making it an extra task they are expected to squeeze in.

For a team of around 10, one champion is a sensible place to start. Make sure the role is about helping colleagues, not pressuring them to use AI for every task. Some processes may not be suitable, and some staff will need more time and support.

Set simple privacy rules before staff begin

People are more likely to use AI confidently when they know what they can and cannot enter. Before training, check the tool’s settings and terms, including how it handles information submitted by users.

For a first trial with a public AI tool, keep the rules simple:

• Use fictional or de-identified examples for practice.
• Don’t paste customer or employee personal details, passwords, confidential documents or sensitive business information into a public tool.
• Don’t use AI as the only basis for important decisions about a person.
• Have a person check any result before it is sent, shared or acted on.

The Office of the Australian Information Commissioner’s guidance on using commercially available AI products explains that privacy obligations can apply to personal information entered into AI and to output that contains personal information. It also recommends care with personal information in publicly available generative AI tools.

Your rules may differ depending on the tool, the data and your business. Make sure staff know who to ask if they are unsure, and what to do if they accidentally share information they should not have.

A one-page training session outline to copy

Use this outline for your first team session. Adjust the example to suit your business.

Time	Activity
0–5 minutes	Explain the aim: test whether AI can help with one everyday task.
5–15 minutes	Cover the approved tool, privacy rules and what information not to enter.
15–25 minutes	Demonstrate one low-risk example, such as drafting a follow-up email from fictional notes.
25–40 minutes	Ask each person to try the same task or adapt it to their role.
40–50 minutes	Review the output together. Check facts, missing details, tone and anything the AI invented.
50–60 minutes	Save the best prompt in the shared library. Agree on one task to test over the next week.

Finish by asking each person: “Would you use this again? What would need to change?” Their answers will help you improve the prompt or choose a better task.

Make the learning part of the work

One training session is a starting point, not the whole plan. Ask staff to use the prompt on the chosen task during the week, then spend 10 minutes discussing what worked and what did not.

Look for practical signs of progress:

• A task takes fewer steps
• A first draft needs fewer changes
• Staff are using the shared prompt
• People know when not to use AI
• Errors are being noticed before work goes out

Don’t measure success by how often staff open a tool. Measure whether it helps with a real piece of work without creating extra checking or risk.

AI also won’t know your business perfectly, make every decision well or take responsibility for a mistake. Keep a person in charge of the final result, especially for customer communications, financial information, staff matters and anything with legal or safety consequences.

Help your team take the next step

If AI is sitting unused in your business, start with one task, one short session and one clear set of rules. Then build on what your team learns.

AI Driven Biz helps businesses put AI into practice and train staff to use it in their day-to-day work. You can also read more about AI and your business.

A low-pressure next step is to ask your team which repetitive task they would most like to make easier, then use that as the focus for your first session.