Artificial Idea | AI careers · practical prompts · no hype Thursday, October 30, 2025 · Issue #26 · Prompt Tutorial
The HR playbook
5 prompts for HR professionals to handle the questions AI can't answer
AI is changing every function in every organisation. HR is changing faster than most, and the professionals inside it who understand what AI can and cannot do are the ones being asked to lead the transition.
Issue #25 made the case that the bifurcation in the labour market is real, measurable, and compounding at a rate that makes engagement with it increasingly urgent. This issue turns to a function that sits at the centre of that bifurcation in every organisation: human resources.
HR is in an unusual position in the AI transition. It is simultaneously one of the functions most affected by AI adoption, responsible for managing the workforce implications of a technology that is restructuring the workforce, and one of the functions where the limits of AI are most consequential. The decisions HR professionals make about people, about fairness, about development, and about organisational culture cannot be delegated to a tool without losing something essential in the delegation.
The HR professionals navigating this well are those who have developed a precise understanding of where AI adds genuine value in their function and where it does not, and who have built workflows that use AI for the former while protecting the human judgment required for the latter. These five prompts are designed for that purpose.
They are also relevant to any professional who needs to understand how HR actually works, because the professionals who understand the function that manages their career trajectory are consistently better positioned than those who treat it as an administrative background process.
Prompt 1: The job architecture analyser
The problem it solves: evaluating whether a role's current job description accurately reflects what the organisation actually needs, which in most organisations it does not, because job descriptions are rarely updated with the frequency that actual role requirements change.
You are a senior HR business partner with
deep experience in job architecture and
role design helping me evaluate whether
a current job description accurately
reflects the role's actual requirements.
Current job description: [paste]
What the role actually involves based
on my knowledge of the function and
the organisation's current priorities:
[describe in as much detail as possible]
The organisational context: [what is
changing in this function, what pressures
the team is under, what the organisation
is trying to achieve in the next
twelve to eighteen months]
Please:
1. Identify the gaps between what the
job description specifies and what
the role actually requires, in order
of significance
2. Identify any requirements in the
job description that are no longer
relevant to what the role actually does
3. Assess whether the current job description
would attract the profile of candidate
most likely to succeed in the role
as it actually exists, and where it
would attract the wrong profile
4. Draft a revised role summary of
three to four sentences that accurately
reflects what the role requires and
what kind of professional would
thrive in it
5. Identify the two or three capabilities
that are currently missing from the
job description that AI-augmented
work in this function now requires
Flag any area where the gap between
the described role and the actual role
is large enough to create a legal or
equity risk in the hiring process.
The legal and equity risk flag is not incidental. Job descriptions that do not accurately reflect role requirements create measurable risks in hiring processes, particularly when the gap between the description and the actual role affects which candidates self-select into the pipeline. HR professionals who conduct this analysis proactively are managing a risk that their organisations often do not know they are carrying.
Prompt 2: The difficult conversation preparer
The problem it solves: preparing for the specific category of HR conversation that is most consequential and most difficult, which is the conversation with an employee whose performance, behaviour, or situation requires a direct, honest, and legally defensible discussion.
You are an experienced HR director helping
me prepare for a difficult employee conversation.
The situation: [describe the performance issue,
behavioural concern, or circumstance requiring
the conversation, including the history,
what has been communicated previously,
and what outcome the conversation needs
to achieve]
The employee context: [their tenure,
their general performance history,
any relevant personal circumstances
you are aware of, and your assessment
of how they are likely to respond]
Organisational context: [any relevant
policies, previous commitments made,
or stakeholders who need to be
kept informed]
Please help me prepare:
1. The opening statement: how to begin
the conversation in a way that is
direct without being aggressive,
and that establishes the purpose
clearly without creating immediate defensiveness
2. The core message: the three things
that must be communicated in this
conversation regardless of how
it develops, stated in plain language
3. The likely responses: the three
most probable reactions from this
employee and how to handle each
one without losing the thread
of what the conversation needs to achieve
4. The documentation: what should be
documented before, during, and
after this conversation to create
an accurate and legally defensible record
5. The follow-up: what needs to happen
within 48 hours of this conversation
to ensure the outcome is properly
supported and tracked
Constraint: this preparation must be
grounded in what is fair and honest
to the employee, not only in what
is legally defensible for the organisation.
Both matter. The preparation should
reflect both.
The constraint that the preparation must reflect what is fair to the employee, not only what is legally defensible for the organisation, is the one that makes this prompt useful for HR professionals who take their function seriously rather than those who treat it as organisational risk management with a human face. The two objectives are compatible in well-run organisations and in tension in poorly run ones. The prompt is designed to serve the former.
Prompt 3: The organisational culture diagnostician
The problem it solves: turning qualitative signals from employee surveys, exit interviews, and informal feedback into a structured diagnostic that identifies the specific cultural dynamics most likely to affect performance, retention, and the organisation's ability to navigate change.
You are an organisational psychologist helping
me diagnose cultural dynamics from qualitative
employee data.
Data sources available: [describe the data:
employee survey results, exit interview themes,
informal feedback, manager observations,
or any combination]
Here is the data: [paste or summarise]
Organisational context: [the organisation's
current situation, any significant changes
underway, and the specific cultural challenges
leadership has identified or suspects]
Please:
1. Identify the three most significant
cultural dynamics visible in this data,
described in terms of the underlying
organisational behaviour rather than
the surface symptoms
2. Identify any pattern in the data that
is likely to be a leading indicator
of a problem that has not yet fully
manifested, and why
3. Identify any gap between what the
organisation believes about its culture
and what the data suggests is
actually true
4. Assess which of the identified dynamics
is most likely to affect the organisation's
ability to navigate the specific changes
currently underway
5. Propose one intervention for the most
significant dynamic that is specific
enough to be implemented within
thirty days and measurable enough
to evaluate within ninety
Flag where the data is insufficient
to support a confident diagnosis and
what additional data collection
would most improve the analysis.
The instruction to identify gaps between what the organisation believes about its culture and what the data suggests is actually true is the most politically sensitive and most analytically valuable component of this prompt. Organisational cultures are frequently misperceived by the people inside them, particularly at senior levels where the filtered nature of upward communication produces a systematically distorted picture. Naming that gap explicitly, with data support, is the diagnostic work that makes cultural interventions possible rather than cosmetic.
Prompt 4: The talent risk identifier
The problem it solves: identifying which employees represent the highest retention risk before they resign, using the patterns in available data to surface leading indicators that are consistently present before voluntary departures but consistently missed until it is too late to address them.
You are a workforce analytics specialist
helping me identify talent retention risks
before they become departures.
Available data about my team or organisation:
[describe what you have access to:
performance review history, promotion
timeline data, compensation relative
to market, engagement survey results,
tenure patterns, or any other relevant signals]
Here is the data or a summary of it:
[paste or describe]
Organisational context: [the current
environment, any recent changes that
might affect retention, and the roles
or individuals where departure would
have the highest organisational impact]
Please:
1. Identify the profile most at risk
of voluntary departure based on
the patterns in this data, described
in terms of the characteristics
and circumstances rather than
individual names
2. Identify the leading indicators
present in this data that research
consistently associates with
pre-departure behaviour
3. Assess which specific circumstances
or changes in the current environment
are most likely to trigger departure
decisions among the at-risk profile
4. Propose three retention interventions
ranked by their likely effectiveness
for this specific profile, with
a rationale for each
5. Identify the intervention that
is most commonly attempted and
least effective for this type
of retention risk, so it can
be deprioritised
Be specific about what the data
supports and what is inference.
Retention risk analysis that
presents inference as certainty
produces interventions that
address the wrong problem.
The instruction to identify the commonly attempted but least effective intervention is the component most likely to produce immediate practical value. Organisations facing retention risk tend to reach for the most visible intervention, typically compensation adjustment, regardless of whether compensation is the actual driver of the risk. Identifying the mismatch between the habitual response and the actual problem is often more valuable than identifying the right response, because it prevents wasted resource allocation before the right response is considered.
Prompt 5: The AI policy drafter
The problem it solves: developing a practical, specific, and enforceable organisational policy for AI tool usage that addresses the real risks without being so restrictive that it prevents the productivity gains AI enables.
You are an HR and employment law specialist
helping me draft an organisational AI
usage policy that is practical, specific,
and legally sound.
Our organisation: [size, industry,
the types of AI tools currently in use
or under consideration, and the specific
risks most relevant to our context]
Our current approach: [describe any
existing guidance, formal or informal,
about AI tool usage]
The specific concerns we need to address:
[data privacy, intellectual property,
accuracy and verification requirements,
equity in AI-assisted decisions,
client confidentiality, or any other
specific concerns relevant to our context]
Please draft a policy framework covering:
1. Permitted and restricted uses:
specific enough that an employee
knows whether a given use is
permitted without having to ask,
but not so prescriptive that
it prevents uses the organisation
would benefit from
2. Data handling requirements:
what categories of information
should not be input into AI tools,
with a clear rationale for each restriction
3. Verification and accountability:
the standard of review required
for AI-assisted outputs before
they are used, communicated externally,
or acted upon, calibrated to
the risk level of different use cases
4. Disclosure requirements: when and
how employees should disclose
AI assistance in their work,
to internal and external stakeholders
5. Review mechanism: how the policy
will be updated as AI tools and
their capabilities evolve,
with a specific review cadence
Constraint: the policy should be
readable by a non-specialist employee
in under ten minutes and actionable
without legal interpretation.
Flag any area where legal advice
specific to our jurisdiction
is required before finalisation.
The constraint that the policy should be readable and actionable without legal interpretation reflects one of the most consistent failure modes in organisational AI policy development. Policies drafted primarily to manage legal risk produce documents that are comprehensive, defensive, and largely ignored by the employees they are meant to guide, because those employees cannot extract from them a clear answer to the practical question they are actually asking, which is: can I use this tool for this purpose?
A policy that answers that question clearly for the most common use cases, while flagging the edge cases that require further guidance, is followed. A policy that requires legal interpretation for every use case is filed.
A note on what these prompts cannot do
These five prompts cover the analytical and preparatory work in HR functions where AI adds genuine value. They do not cover the work where AI does not belong.
The decision to terminate an employee cannot be delegated to an AI analysis. The judgment about whether an organisational culture is healthy requires human presence and relationship that no data summary can substitute for. The conversation with an employee in distress requires a human being, not a prepared script. The assessment of whether a candidate is right for a role involves dimensions of judgment, intuition, and interpersonal reading that sit outside what these tools can produce.
The HR professionals who use AI most effectively are precise about this boundary. They use the tools to do the analytical groundwork faster and more thoroughly, freeing the time and cognitive capacity for the human work that the analysis is meant to support. The boundary between the two is not always obvious and it shifts as the tools develop. Maintaining clarity about where it currently sits is part of the professional responsibility of anyone using these tools in a function where the consequences of getting it wrong are borne by people rather than processes.
Monday we are examining the most counterintuitive finding to emerge from the AI and careers research this year: the characteristic most correlated with successful navigation of the AI transition is not technical skill, not formal education, and not seniority. It is something simpler, less glamorous, and more within the reach of every professional reading this than any of those variables.
The finding is well-supported and almost entirely absent from the mainstream AI career conversation. Monday's issue puts it where it belongs.
— The Artificial Idea team

