Make the task clear
Ask your HR and payroll teams to walk through it together. List the information they receive. Identify the checks they perform. Name the person responsible for each approval and the final record.
Start with work your team already knows. An attendance correction is a useful example.
Follow one attendance correction
Imagine an employee asking to correct a missed time-in before payroll.
What does the employee submit? Which attendance record does HR check? Who confirms what happened? Who approves the correction? How does the approved change reach payroll?
Follow the request from submission to completion. Include the message sent back to the employee.
Then examine the exceptions. The approver is absent. The request arrives after the cutoff. The employee worked at another branch that day.
For a food group, BPO or manpower business with several companies and locations, these details deserve attention. Confirm the correct employee and employer. Identify the branch or assignment involved. Check who may access the records.
Your team should be able to explain each step before asking a tool to help with it.
Five questions to ask together
- What starts this task and what outcome do we need?
- Which employee, company and branch does it involve?
- What records do we use and how do we check their accuracy?
- Who reviews the request and handles approvals or exceptions?
- How do we confirm completion and record the decision?
Ask the people who do the work every day. A written procedure may leave out the calls and follow-ups needed to finish a request.
Put those steps on one page. Look for missing information and unclear approvals. Check whether two people repeat the same task. Agree on which record the team should rely on.
You may find useful improvements before introducing AI.
Run a small exercise
Use a tool approved by your company. Create a fully fictional attendance-correction example. Provide a short process description without employee information or confidential company details.
Ask AI to suggest a checklist or identify missing steps. Treat its response as a draft.
Compare every suggestion with your approved process. Check whether it added a rule your company does not have. Test the checklist against the exceptions you identified.
Record what helped and what needed correction. This gives your team something concrete to discuss before expanding the exercise.
Keep real employee records out of this first test. Before considering their use, involve the people responsible for privacy and security.
The National Privacy Commission’s AI guidance emphasizes accurate data and accountability. It also addresses mechanisms for human review.
National Privacy Commission Advisory No. 2024-04
For an attendance correction that affects pay, keep a responsible person in charge of the final decision. Tell employees whom to approach if something is wrong.
Choose a clear next step
After the exercise, decide whether AI helped with that specific task. Consider the effort needed to check and correct its output.
Your team can then decide what to test next. It can also decide that the process needs more work first.
Preparing HR for AI begins with understanding the work well enough to judge where help is useful.
Book an OLi HR demo for your team.
Original practical guidance prepared for OLi with AI assistance. It focuses on workflow and product evaluation. It does not replace your organization’s policies or professional advice.
