HR process automation removes the repetitive work in onboarding, offboarding, documents, holidays and payroll variables, without touching decisions about people. The team stops copying data and chasing signatures and gains time for work that needs judgement. Because employee data is involved, it is wise to decide from the start where it is processed and who can access it.
Which HR processes can be automated
People processes carry a lot of administrative work that repeats every week. These are the most common:
- Onboarding. Creating accounts, granting access, preparing the contract and welcome documents.
- Offboarding. Removing access, recovering equipment, preparing the final settlement and filing the record.
- Documents. Requesting, receiving, checking and filing certificates, supporting paperwork and contracts.
- Holidays and absences. Requesting them, approving them and reflecting them in the team calendar.
- Payroll variables. Collecting overtime, incidents and absences and sending them to payroll without retyping.
- Internal questions. Answering frequent questions about policies and procedures.
They all share three traits: high volume, clear rules and data that currently travels by email and spreadsheet.
They also involve several departments at once. When each works with its own tool, people end up acting as the bridge between systems, and that is where things get forgotten or delayed.
Why onboarding and offboarding are the best place to start
Onboarding involves several departments: HR, IT, finance and the new hire’s manager. When it is coordinated by email, it is easy for an account, an access right or a laptop to be missing on day one.
Offboarding is even more sensitive. A poorly handled departure can leave access to systems and data open for weeks. It is one of the most common and least visible risks of manual processes.
Once automated, onboarding and offboarding always follow the same sequence:
- The system detects the joiner or leaver in the HR tool.
- It sends tasks to each department, with deadlines and reminders.
- It creates or removes access according to the role.
- It records every step, with date and owner.
The result: everything is ready on day one, and access is removed on the last day.
How holidays and payroll are automated
Holiday requests and payroll variables are usually the processes with the most back-and-forth email.
| Task | Today, by hand | Automated |
|---|---|---|
| Request holiday | Email to the manager | Request in a workflow, with the remaining balance visible |
| Approve | Email reply, sometimes late | One-click approval, with a reminder if delayed |
| Update the calendar | Someone notes it down | It updates itself |
| Send variables to payroll | Spreadsheet checked by hand | Variables reach payroll already validated |
Example: if 200 variables a month go to payroll across overtime, absences and incidents, and each one is checked and typed in, the savings in time and errors show from the first close. The real figure depends on each company and is measured.
What role AI agents play in HR
AI agents are useful for two specific tasks:
- Reading documents, such as certificates or supporting paperwork, and extracting the data to be recorded.
- Answering internal questions about holidays, leave or procedures, using the company’s documentation and citing the source.
In both cases the agent works within clear limits and hands any sensitive question to a person. For how they work and where their limits lie, see our article on AI agents in the enterprise.
What should not be automated
HR automation has a clear limit: decisions about people. Recruiting, appraising, resolving a conflict or communicating a dismissal require judgement and a human touch. Automation prepares the information and removes the paperwork, but it does not decide.
Nor should you automate a process the company is in the middle of redefining. Sort it out first, then automate it.
What to look after with employee data
Employee data is personal data protected by the GDPR, and some of it, such as health data, has stronger protection. Before automating, define:
- Which data each process needs, and process nothing more.
- Who can access each item and with which permissions.
- Where it is stored and processed: on the company’s infrastructure or the provider’s, in line with its security policy.
- How long it is kept and how it is deleted.
Our article on AI and GDPR covers these points in more depth. This information is general and does not replace legal advice.
How results are measured
In HR, improvement is not always measured in hours. It also shows in errors avoided and deadlines met. That is why it pays to agree with the team which indicators matter before changing anything:
- Time per joiner and per leaver, from notification until everything is ready.
- Access still to be removed after a departure, and how long it takes to close.
- Payroll errors caused by variables passed on incorrectly.
- Response time for holiday requests and internal questions.
With the starting point measured, after is compared with before and shared with management. It is the same yardstick we use in any project, as we explain in our piece on how to measure automation ROI.
Where to start
The sensible approach is to begin with the process that takes the most time today and measure its starting point: how many joiners, leavers or requests there are each month and how long each one takes. With that baseline, before can be compared with after. Our use cases include other examples from the people area, and automation integrates with the tools your team already uses.
If you would like to know which of your HR processes could run by themselves, let us start with a first conversation to get to know your company and see where we can help.