Purpose and fit

Use this when a team needs practical development goals. Focus on work the role actually performs and on the support needed to perform it well. Keep developmental practice separate from high-stakes employment decisions.

Before you start

Use current role descriptions, ordinary task examples and approved tool/policy notes. Role-level information is sufficient; you do not need employee profiles or invented performance scores. If no observed work is available, adapt the output to proposed practice activities and mark current proficiency unknown.

Use only information you are authorised to share with the chosen AI tool. Adjust the prompt to the fields and context you actually have; a specially created dataset is not required.

Required inputs

  • Role responsibilities and common tasks, confirmed by role holders
  • Approved tools, relevant policies and expected quality checks
  • Examples of good work and common errors, with appropriate permission
  • Available learning time, accessibility needs and sources of support

Practical workflow

  1. Choose a few representative tasks for the role. Describe what acceptable work looks like, including when AI use is inappropriate.
  2. Identify the skills each task requires: framing the request, selecting appropriate inputs, verifying outputs, handling exceptions and explaining decisions.
  3. Invite role holders to describe their experience and support needs. Use a safe practice task to observe the work, rather than inferring ability from job title or background.
  4. For each priority gap, define one practice activity, the support available and an observable sign of progress.
  5. Review the completed work with the learner. Adjust the plan using evidence and feedback, including changes needed in tools or workflow.

Expected output

Expect a role-level learning plan for discussion with role holders, not an assessment of an individual or evidence for an employment decision.

A role capability matrix and a short learning plan:

Role/task | skill required | observed evidence or unknown | practice activity | support/coach | success check | review date

Useful observable checks include:

  • Can the person identify missing or unsuitable inputs?
  • Can they verify a material claim against the actual source?
  • Can they explain what requires approval and stop at that boundary?
  • Can they recover using the agreed manual process?

Human checks

Role holders validate the task map and can correct inaccurate assumptions. A qualified reviewer examines actual work, using consistent criteria and appropriate accommodations. Keep individual learning records limited to the people who need them. Do not upload personal performance information to an unapproved AI tool or infer capability from demographic traits.

Limitations

A practice exercise samples a narrow set of skills. Tool access, time, process quality and support affect performance. Completing training or producing a fluent prompt does not establish readiness for every task. This plan should not be used for automated employee ranking or selection.

Starter prompt

Draft a development plan for the role [role] using these approved task descriptions: [tasks]. Available tools and boundaries are [tools/rules]. Learning support available is [support]. Use role-level information, not individual employee profiles.

For each task, identify practical AI-related skills, a safe practice activity and an observable check of good work. Include source verification, data handling, approval boundaries and fallback skills where relevant. Mark skill levels unknown unless actual evidence is supplied. Suggest questions to ask role holders before finalising the plan. Do not rank people, infer ability from personal characteristics or recommend employment decisions.

Return one row per supplied role task with the relevant skill, existing evidence or unknown, safe practice using available work examples, support needed and an observable success check. Distinguish proposed activities from observed proficiency and ask about missing support or accessibility constraints. Do not infer individual ability.

Before drafting, ask a focused question if a missing detail materially changes the task or acceptance criteria. Otherwise produce a clearly limited draft and list the gaps. Label any non-critical assumptions; never invent facts, figures, sources, approvals or completed actions.

Data and AI output disclaimer

Use relevant data that exists or that you provide, with accuracy and completeness sufficient for the task. Different AI models and versions may produce different results, and responses may vary between runs. You are responsible for checking and validating the output against your own requirements, source information and acceptance criteria before relying on it.