AI in HR: start with one useful task.

A useful starting point for AI in a people team is a task you can describe clearly. Preparing a first draft of a role brief, organizing approved research or comparing a market map with agreed requirements gives you something specific to evaluate.

Choose one task. Describe what a good result looks like and decide how you will check it. You can then assess whether the tool improves the work and whether it is worth using again.

Set a clear starting point

Record how your team handles the task today. How long does it take? Where do errors or repeated work occur? Who reviews the result, and what do they need to trust it?

For a talent research brief, useful measures might include the accuracy of the information, the relevance of the people identified and the time needed to verify the output. Count the review time as part of the work.

Test the result against the source

• Use information the tool is approved to handle. Agree with the people responsible for your organization’s data which materials can be used.
• Keep the supporting evidence. A reviewer should be able to trace a career detail, quotation or number to its source.
• Check what is missing. A confident answer may overlook relevant experience or infer something the source does not establish.
• Compare the whole task. Assess the quality of the finished work and the time spent correcting it before expanding the use.

Separate career data from personal intent

Career information can help identify experience worth exploring. It does not confirm that a person is interested in a role, willing to relocate or available at a particular level of compensation. Those questions require a conversation.

Keep that distinction visible in a market map or candidate brief. State what is supported, when it was checked and what still needs to be confirmed.

Decide what to do next

After the trial, review the result with the people who use it. Keep the approach if it helps, refine it if the gaps are manageable, or stop if the checking effort outweighs the benefit.

Start with a task your team understands well enough to judge. A clear result from that work gives you a better basis for deciding where AI can help next.

Editorial update, September 23, 2026: removed an adoption-score comparison and personal anecdote that were not supported by a source record.