Tell your people as early and as specifically as possible which tasks AI will take over, what that will change in their roles and what they will do instead. Don’t promise more than you can deliver, and admit what you don’t know yet. Fear doesn’t come from AI itself but from silence, which people fill with the headlines from the news.
Exactly how to say it depends mainly on what you actually plan to do. The conversation is different in a company that wants to use the freed-up time to serve more customers than in one where a role really will shrink. Trust within the team matters too: whether people have experienced management telling them the truth even when it is uncomfortable.
Your people take news of AI layoffs personally
Hospodářské noviny reports that Microsoft is offering voluntary severance to roughly sixty thousand employees. The reason is cost-cutting amid heavy investment in AI. Slovak media, meanwhile, are discussing how Barclays is cutting five thousand jobs, with artificial intelligence being the main reason, according to them.
It isn’t only managers at big companies who read news like this. So do the accountant, the warehouse worker and the sales assistant at a twenty-person company. And when, in the same week, they hear that “we’re trying out AI for invoices”, they connect the dots themselves.
For a small company, this is more serious than it seems. Afraid for their jobs, people stop sharing how they really do their work. They feel they would be helping to replace themselves. Trying out new tools slows down.
And the first to start looking elsewhere are the capable people who have somewhere to go.
Silence is not a neutral choice here. When management says nothing, the team finds its own explanation. Usually a worse one than reality.
Before you speak, be clear about exactly what is going to change
People can tell when management answers in generalities because it doesn’t know itself. So first answer a few questions for yourself. It helps to write them down on a single page.
- Which specific tasks do you want to automate? Not “customer support”, but something like “sorting incoming enquiries and drafting replies to the recurring ones”.
- Who does these tasks today, and what share of their work do they make up? An hour a week is a different situation from most of someone’s working time.
- What will happen to the freed-up time? More customers for the same team, faster responses, work there was never time for, or less need for people in one role.
- What don’t you know yet? For example, whether the tool will work reliably enough. That’s fine, you just need to be able to say it out loud.
- What can you promise and keep? Dates, training, involving people in testing, information about the next step.
If the answer to the third question is “we want fewer people”, don’t hide it behind talk of growth. People will see through it anyway. You will then lose the trust even of those the change doesn’t affect at all.
Talk about tasks, not people
The sentence “AI will replace the invoicing clerk” sounds like a verdict. Yet it usually isn’t true. AI takes over some tasks, not a person’s whole job.
So break each affected role down into tasks and show what changes where. A simple table helps, and you can show it to your people too:
| Role | What automation takes over | What stays with a person | What may be added |
|---|---|---|---|
| Accounts payable | Reading received invoices and entering them | Checking disputed items, communicating with suppliers | Monitoring payments, cost analysis |
| Sales support | Copying orders from e-mail into the system | Exceptions, non-standard orders, customer care | More time for customers who buy less than they could |
| Customer support | Drafting replies to recurring enquiries | Approving replies, complex and sensitive cases | Suggestions for improving the service |
| Warehouse and shipping | Order status messages to customers | Packing, checks, handling complaints | Adjusting processes where errors occur |
The table is an example, not a template. Your tasks will be different.
What matters is that people see it for their own role and can say what is missing. You will often find that a task that looks simple has exceptions only the person doing it knows about.
Tell everyone at once, then each person individually
A sensible approach has five steps:
- A short all-hands meeting. Say why you are introducing AI, which tasks it affects, what isn’t changing for now and when you will know more. Make sure everyone hears it in the same words and from you, not through the grapevine.
- Conversations with the people most affected by the change. The same day or the next. Go through each person’s role using the table and ask what they would like to do differently.
- A written summary. A few paragraphs by e-mail or on the noticeboard: what is changing, what isn’t, who people can ask about what. Under stress, people remember little and then retell it at home in their own way.
- Space for questions without an audience. Not everyone will ask in front of colleagues whether they are going to lose their job. Let them know they can drop by or get in touch in writing.
- A date for the next update. Say specifically when you will return to the topic, for example a month after the first trial. And keep that date, even if there is nothing new to say.
Answer the direct question “am I going to lose my job?” directly. An evasive answer is heard as a yes.
Only promise what you will actually deliver
A common mistake is a reassuring promise that no longer holds six months later. “Nobody here will ever lose their job because of AI” sounds good. But nobody knows what the company will look like in a few years.
Promises that are bounded and verifiable are better:
- “We are not cutting any positions because of automation before the end of the year.” Only if that is really the case.
- “Before anything changes in your role, you will hear it from me well in advance.”
- “Training on the new tools takes place during working hours and the company pays for it.”
- “We will decide together with you how the task will be done from now on.”
If a role really will shrink or disappear, tell the person affected among the first and in private. Then talk about the options: other work within the company, retraining, time to look for a new job, a reference.
What exactly you must and may do when changing or terminating employment differs from country to country. Before you announce anything, check with an employment law specialist who knows the local rules.
People who see the company act fairly even towards someone whose job was eliminated trust it more for themselves too.
Instead of uncertainty, give people a role in the change
Fear shrinks most when people feel they are shaping the change themselves rather than having it happen to them. And the people who do a task every day know where automation will help and where it will make mistakes.
Specifically, you can offer:
- Involvement in testing. Let the people it affects try out the new tool and say where it fails. Take their feedback seriously and show what you changed because of it.
- Ownership of the outcome. The person who used to key in invoices can check whether the automation enters them correctly and handle the exceptions. Their knowledge becomes more valuable, not redundant.
- Learning during working hours. Time to try AI tools in their own work, not an extra obligation in the evenings.
- A clear plan for the freed-up time. Say what work you want to use the time saved by automation for. After a few months, show whether that is really happening.
Check-in questions after the first month
- Do people ask about the new tools on their own, or do they avoid the topic?
- Do they report automation errors, or quietly work around them?
- Has someone you are counting on unexpectedly started asking about other jobs?
- Does everyone in the affected roles know what they will be doing six months from now?
If your answer to any of these questions is unclear, go back to the conversations. Fear doesn’t subside after one meeting. It subsides when people see, again and again, that what you said holds true.
Sources
Frequently asked questions
Should I talk about introducing AI even if I don’t know yet how it will turn out?
Yes, and sooner rather than later. Say what you are trying, why and when you will know more. Uncertainty acknowledged by management is easier to bear than uncertainty people piece together from hints.
What if someone asks me directly whether they will lose their job and I don’t know?
Say that you don’t know yet and explain what the decision depends on. Add when you will know more and that they will hear it from you first. Vague reassurance along the lines of “don’t worry” sounds more like confirmation of their fears.
Should I offer people a share of what automation brings in?
It can help if the benefit is measurable and the rules are simple. But a reward won’t replace clear information about what will happen to their role. Before introducing one, check with a local specialist how to set it up correctly within your pay policy.
What should I do if someone the company needs leaves after the announcement?
Ask them openly about the reason, for example in an exit interview. If it was fear of automation, your message probably didn’t land the way you intended. Go through what is changing with the team again and find out whether anyone else has similar concerns.
Should I bring up the news about layoffs at big companies myself?
Yes, by all means, people know about it anyway. Explain how your situation differs and where it perhaps doesn’t. If you skip over it, it will look as though you are avoiding it precisely because it applies to you.