Every few months there is a new promise that artificial intelligence will transform HR: faster hiring, instant answers to employee questions, automatic reports. Some of it is real. Much of it assumes a foundation that many companies do not have.
Speed is not direction
A tool applies itself to the process it finds. If the process is clear, with defined owners, consistent rules and reliable data, technology can remove effort and delay. If the process is unclear, the tool reproduces the ambiguity, and does so quickly and at scale.
Consider the common case of a company that wants a system to answer employee questions about leave and benefits. If the policies themselves contradict each other, or exist in several versions across managers' inboxes, the system will give inconsistent answers with great confidence. The problem was never the speed of response. It was that the company had not settled what its policy was.
A faster answer to an unsettled question is still an unsettled answer.
What to settle first
Three things come before tools.
Decision rights. For each recurring people decision, such as hiring, pay changes, promotion and termination, who proposes, who approves, and on what basis? If this lives in people's heads, automation has nothing to follow.
One source of employee data. Role, grade, reporting line, pay and contract terms should exist in one trusted place. When the same fact lives in three spreadsheets, any report or any model built on them inherits the disagreement.
A named owner for each process. Somebody has to be accountable when the process fails. Without that person, a new system becomes an extra place for work to get stuck.
Where AI can help
Once those foundations exist, AI has useful work to do. It can draft job descriptions for a person to edit, summarise written feedback so that a manager can see themes, answer routine questions from a clean and current policy base, and prepare first versions of routine documents. In each case the person remains responsible for the decision, and the tool saves time on the preparation.
It is a poor fit for decisions that need judgement about people, such as who to promote or whether a performance concern is fair, and for any use where the data is unreliable. It also raises a question that deserves attention from the start: employee information is personal data, and where it is processed and who can see it should be decided deliberately, not by whichever feature was switched on.
The data problem is also a trust problem
Employees notice quickly when a system gives wrong answers about their own pay, leave or contract. One visible error costs more trust than the system saved in time. That is why clean data comes before automation. The people who will use the tool, whether employees asking questions or managers reading reports, need to see that what it tells them matches what they know to be true. Trust built slowly through accurate answers is easy to lose to a handful of confident mistakes.
What this means for smaller companies
Smaller companies sometimes assume the foundation work is for large organisations with big HR teams. The opposite is closer to the truth. A company with a few hundred people can settle decision rights and clean its data in weeks, because there are fewer people to align. The benefit also arrives sooner, since there is less legacy to untangle. The choice of tool then becomes a modest decision, made on a clear brief, instead of a large bet made in the hope that software will impose order that the company has not yet defined.
A test before you buy
Before buying or building anything, try to draw the process on a single page: the steps, who does each, what data is used and who decides. If the page cannot be drawn, or if people disagree about it, that is the work to do first. If it can be drawn, the choice of tool becomes simpler, because you know what the tool has to do.
Order is the strategy
The companies that benefit most from technology in HR tend to be those that did the dull work first. The sequence of structure, then policy, then data, then tools is not the exciting part. It is the part that decides whether the exciting part works.
Thinking about this in your own business?
Explore HR Technology