People policy assistant
Employee / manager question → approved policy → grounded answer + escalation.
Return people teams to the work that needs people.
HR teams spend large amounts of time moving information, answering repeat questions, preparing documents and coordinating processes. AI can reduce that load while preserving the human accountability required for employment decisions.
FN/
AUSTRALIA / APPLIED AIRecruitment, onboarding, policy, learning, employee questions and manager support all create recurring administrative work. The useful line is clear: automate preparation and retrieval, not human judgement about people.
Buyer: CPO · Head of People · HR Director
01 / WHERE TO START
Three workflows worth testing first.
These are not universal prescriptions. They are practical starting patterns with a clear owner, observable work and a measurable outcome.
Onboarding coordination
New starter → role / location → checklist → owners → status.
Recruitment administration
Role + applications → structured comparison inputs → human selection.
02 / 12 WORKFLOWS
Go deeper than a list of AI tools.
Every workflow is framed as work moving from input to action. That is what lets us map systems, context, human review and the business measure behind it.
Policy retrieval
Question → policy → answer + sourceOnboarding workflow
Starter → tasks → owners → statusRecruitment administration
Applications → structured summaryInterview preparation
Role + candidate context → guideLearning support
Role / need → approved learning pathManager support
People process → steps / templatesEmployee query triage
Question → category → response / routeDocument preparation
Approved templates → first draftWorkforce reporting
People data → trend → narrativeExit administration
Trigger → checklist → ownersRole / skills mapping
Role library → capability viewHR knowledge maintenance
Policy changes → knowledge updatesThe workflow is the product.
A useful implementation defines five things before it ships: the trigger, the business context, what AI is allowed to do, where a person owns the decision, and what happens when the normal path fails.
TRIGGER→CONTEXT→AI ACTION→HUMAN OWNERSHIP→EXCEPTION→MEASURE
04 / WHAT TO MEASURE
AI usage is not the outcome.
Use operating measures that already matter to the function, then compare the new workflow against a baseline.
01HR admin hours
02Time to onboard
03Query response time
04Manager self-service
05Process completion
06Recruitment cycle time
07Policy search time
08Adoption
Keep accountability visible.
Automation should make the operating boundary clearer, not hide it.
Hiring / termination decisionsPerformance decisions without accountable human reviewSensitive employee profilingUse of protected attributes for automated decision making
06 / FIRST 90 DAYS
Map. Build. Activate.
Map
Choose the business outcome. Baseline the current workflow. Map systems, data, handoffs, exceptions and ownership. Rank the first opportunities by value and feasibility.
OUTPUT: first implementation briefBuild
Design against real context and systems. Define permissions, human review, governance and exception paths. Test with the people who run the work.
OUTPUT: production-ready workflowActivate
Launch, train, measure, fix the operating design and capture the reusable pattern. Use the evidence to choose the next workflow.
OUTPUT: live workflow + outcome evidenceShow me what AI could change in my department.
Select your team, size and pain point. Identify will generate five workflow opportunities, the impact dimensions to test and a recommended starting point.
RUN THE PEOPLE & HR FINDERFORTNIGHTLY AI BRIEFINGWhat matters in applied AI.
What matters in applied AI.
Without the noise.
Practical AI workflows, implementation lessons, Australian market signals, governance notes and upcoming Identify events for leadership teams.
Where should People & HR start with AI?
Start with a repeated workflow that creates visible friction, has a clear owner and can be measured before and after. The strongest first move is usually not the most ambitious use case. It is the one that proves the organisation can Map, Build and Activate a useful change.
Do we need to replace our existing systems?
Usually not. Identify designs around the systems, data, permissions and workflow already in the business. The first question is where AI should sit in the operating model, not which platform should replace everything.
What should we measure?
Use the measures already used to run the function: time, cycle time, throughput, service, conversion, quality, exceptions, cost to serve or another agreed business outcome. AI usage on its own is not the result.
What should AI not own?
Material decisions, sensitive judgement and exceptions need clear human accountability. The exact boundary depends on the workflow, risk and industry.