6 min read. Updated 20 October 2026

Copilot training for NZ government teams

Confirm agency approval, the account environment, permitted information and the roles responsible for reviewing outputs. Select fictional or cleared scenarios and test the required features before the workshop. Use public-service AI guidance alongside agency policy. Training develops practice; it does not grant procurement, security, privacy or operational approval.

Read the guide’s sources.

See Copilot training NZ, Copilot courses NZ and Copilot workshops NZ.

Illustration generated with AI.

What should NZ government teams confirm before Copilot training?

Confirm agency approval, the account environment, permitted information and the roles responsible for reviewing outputs. Select fictional or cleared scenarios and test the required features before the workshop. Use public-service AI guidance alongside agency policy1. Training develops practice; it does not grant procurement, security, privacy or operational approval.

Key points

  • Use agency decisions to define the exercise boundary.
  • Test account features before the session.
  • Keep training evidence separate from operational approval.

What problem should the workshop address?

Begin with a work task and the people accountable for it. An agency might want staff to prepare clearer internal summaries, identify unanswered questions in a service brief or check a draft action list. A broad request to teach Copilot leaves the trainer guessing about both the information that can be used and the standard the output must meet.

Ask managers and participants to describe where review effort is spent today. The exercise should help them practise that work, not merely demonstrate a product feature. Choose something that can be represented with public or invented information so participants can discuss errors without exposing a live case or sensitive policy matter.

The public-service responsible-AI guidance provides a framework for examining organisational use1. Apply the agency’s own decisions to the proposed session. A workshop booking is not evidence that an account, integration or new operational process has passed the agency’s required approval steps.

Who needs to help prepare the session?

Include the learning organiser, a business owner for the task and someone who can confirm the Microsoft environment. Where information-handling questions arise, involve the relevant privacy, security or records staff. They do not all need to attend every exercise, but the unresolved decisions should have an owner before delivery is agreed.

The business owner should describe what staff may do with a draft and who approves the final output. The account administrator should confirm which experiences are available and whether the planned app features can be used. The organiser can then give participants clear preparation instructions instead of asking them to bring whatever documents happen to be nearby.

Record practical access needs as well. Staff may use different devices or require an alternative way to participate. Ask about those requirements through the agency’s normal process and prepare an equivalent text-based exercise where appropriate. The objective is to practise the same reasoning and review steps, not to make everyone follow an identical sequence of clicks.

How do we choose an approved practice pack?

Use a fictional service scenario with enough detail to be useful and enough gaps to require judgment. For example, invent a public information page whose opening hours are under review. Supply a current statement, a proposed change and an internal note saying that approval is pending. No real service or agency needs to be named.

Ask participants to separate information that may be stated now from information requiring confirmation. The draft should not publish the proposed hours as settled. Include a question about who approves the change and where the authoritative page will be maintained. These details make the exercise about public-service work rather than generic email writing.

Check the files before distribution. Remove real names, copied correspondence and comments that were not intended for the pack. Label all invented records clearly, and explain that the scenario is a training construction. If a real public document is used instead, retain its source and date so participants can inspect the actual context.

What account and licence checks are necessary?

Microsoft distinguishes Copilot experiences according to account entitlement and the features being used2. Ask the agency’s administrator to verify the planned exercise in the intended environment. Do not infer availability from a consumer demonstration, another organisation’s screenshot or a general statement that staff have Microsoft accounts.

Prepare a short list of tasks to test: opening the approved file, using the intended app experience and retaining the draft in the permitted location. Record what is supported and what needs an alternative. A missing feature should lead to a revised exercise or a clear support question, rather than pressure to change an agency control during training.

Keep additional purchasing separate from the learning plan. A workshop can use invented text to teach prompting and review while access questions remain unresolved. Any proposal to buy licences, enable connectors or expand permissions belongs with the agency’s authorised decision makers and processes.

What should the workshop agenda look like?

Open with the agreed purpose and information boundary, then demonstrate the fictional task and its review standard. Give participants time to inspect the source before generating a draft. Ask them to identify unsupported statements and missing questions, revise the prompt, and compare the next answer with the original material.

Use paired review so participants experience checking someone else’s output. One person drafts the service-page summary while another traces its claims to the source pack. Swap roles with a different example. This makes the checking effort visible and exposes instructions that depend on unwritten knowledge held by the original drafter.

Finish by documenting the method and unresolved issues. Keep a checked example, a flawed example and the steps that distinguished them. Do not end with a general declaration that the tool is ready for live work. The agency can use the exercise evidence to decide what further preparation or approval a bounded pilot would require.

Agenda stagePracticeEvidence retained
ScopeIdentify permitted inputs and output useAgreed exercise boundary
DraftSummarise the fictional service changePrompt and first answer
ReviewTrace claims and preserve pending approvalCorrections with source references
HandoverRepeat with another participantChecked example and open questions

How should records and escalation be handled?

NZ Digital government’s AI records guidance directs agencies to consider the information and records created by AI-supported work and consult authoritative records advice3. Ask the agency’s records team what the actual workflow should retain. The training session should follow that decision rather than invent a retention schedule or assume chat history is the official record.

For an invented exercise, retain the source pack, prompts and review notes in the agreed training location. Clearly distinguish these materials from operational advice. If participants later adapt the method to real work, the responsible owner should decide what evidence of review and approval needs to accompany the accepted output.

Provide a contact for access problems and information concerns. Participants should know when to pause and whom to ask. An uncertain output is a reason to inspect the source, while an inappropriate input may require the agency’s separate incident process. The workshop should help staff recognise the distinction without trying to replace that process.

What evidence would justify a small pilot?

Choose a task with a clear acceptance standard and collect observations on quality and total effort. Include the time taken to check and repair the draft, not just generate it. Ask whether another participant can repeat the method with a different permitted example. A single successful demonstration provides only a narrow basis for a decision.

Document failures as well as useful results. If the assistant treats a proposal as an approved decision, record how that error was detected and what oversight would be needed in real work. If the source is ambiguous, the next improvement may be to the information process rather than the prompt. Keep those conclusions distinct.

The agency may decide to revise, limit, continue or stop the pilot. Record the reasoning and the owner of the next action. Training can prepare people to run that evaluation, but it should not be marketed as proof of compliance, security assurance or a guaranteed efficiency gain.

How should a buyer request a training proposal?

Send a brief describing the staff roles, intended tasks, approved account environment and preferred delivery. Identify any constraints on material, recordings, access or remote participation. Explain who will review the outputs and what you want participants to be able to practise after the session. This gives the provider a concrete basis for scope.

Compare the proposed preparation, exercises and follow-up with those needs. A general product introduction may be sufficient for some staff, while a group responsible for detailed review may need more practice. Avoid assuming that a long agenda is better; the available time should allow participants to inspect and correct their work.

The related Copilot courses provide starting points for foundations, manager review and rollout planning. thecolab.ai can discuss a private session using invented or cleared scenarios. The agency retains its own purchasing and operational approvals, and any public workshop date should be checked on the workshops page before it is treated as available.

Related

  1. 1Introductory

    Microsoft 365 Copilot foundations workshop

    A half-day of Microsoft 365 Copilot training: check your access, draft an email, review a document and test a spreadsheet answer using cleared examples.

  2. 2Leadership

    Copilot for managers

    Help managers catch up, brief from files they can open, and set a checking rule for the team.

  3. 3Organisation

    Copilot rollout and readiness

    Prepare the tenant, choose a pilot, and measure whether people use Copilot well.

Industries and teams

More on office team practice

See all guides.

Questions

What should NZ government teams confirm before Copilot training?

Confirm agency approval, the account environment, permitted information and the roles responsible for reviewing outputs. Select fictional or cleared scenarios and test the required features before the workshop. Use public-service AI guidance alongside agency policy1. Training develops practice; it does not grant procurement, security, privacy or operational approval.

Train your team

  1. ScopeHeadcount, sites, Microsoft 365 plan and licences, and the teams you want to start with.
  2. PlanWe propose a course, a pilot group and a format, and agree the files and meetings to use.
  3. TrainOn-site or over Teams, in Outlook, Word, Excel and Teams, with your own work.
  • In person
  • Remote
  • Half-day workshop
  • Full-day workshop
  • Team rollout

Independent training by thecolab.ai. Not affiliated with Microsoft.

Enquire

Tell us about your team. We reply by email with a suggested course and format.

Not sure which licence? We check before training.

Include the apps, staff roles, number of people and location you have in mind.

Team workshops from $4,000 + GST for up to 10 people. +$200 + GST per extra person.

Public sessions from $399 + GST per person. Schedule coming soon.

No payment now. We reply by email with a suggested course and format.

We use these details only to reply to you, and store them securely with thecolab.ai. Privacy

Sources