How to announce an AI rollout without it sounding like a redundancy warning

Why “greater productivity” sometimes sounds like “fewer jobs”

The CEO opens the meeting with: “Thanks to artificial intelligence, we’ll be producing reports 40% faster.” The slide shows the savings, and the room falls silent. A finance specialist asks the colleague next to her whether cuts are on the way. The team leader doesn’t know the answer. At that moment, it no longer matters whether the presentation looked polished. The company has sparked a conversation about job security, even though it meant to talk about technology.

There is no need to assume that every employee is automatically afraid of losing their job. Some ask about data confidentiality, others about quality, and others still about whether their experience will be valued. So the board should explain the specific changes and their possible effects, rather than assume everyone will understand the message in the same way. If the project is about improving customer service, you need to explain which activities will change, who will judge whether the output is correct and which skills will still be needed. A savings figure on its own doesn’t answer these questions.

Trust grows when the announcement answers the questions people are really asking

A 2025 Pew Research Center survey of US workers showed that concerns about the future of AI were more common than enthusiasm. It does not measure the views of Polish employees, nor does it prove that every company will meet identical reactions. It is a reminder, however, that the board cannot equate familiarity with the technology with a sense of security. Research on communication after mergers, meanwhile, suggests that realistic, repeated information about change can reduce some of the negative effects of uncertainty. A finding from mergers is not a ready-made recipe for an AI rollout, but it supports a sensible principle: don’t replace facts with promotional slogans.

A good announcement should contain four answers: why the company is introducing the tool, what will change in people’s work, what the limits on its use are and what has not yet been decided. Before sending the message, it is worth reading it from the perspective of the person doing the work, not just the person calculating the return on investment. A sentence such as “we will gain the ability to scale our operations” can be replaced with an example: “The tool will prepare a first draft of the reply, and the employee will check that it matches what was agreed with the client.”

Prepare managers first, then announce the change

The biggest problem arises when an employee learns about a new policy during a livestream of the board’s address and their manager is seeing it for the first time too. It is the manager who will get the questions about tasks, development, headcount and how performance will be assessed. If they don’t know the answers, they will improvise or refer everyone to the recording. That is how departments start to receive conflicting information, even when the central announcement was carefully prepared.

The board should brief managers beforehand and give them all the same set of facts, questions and boundaries of responsibility. The point is not to teach everyone the same slogans. Managers need to understand when they can answer themselves, when to pass a topic on to a specialist and whom to tell about a problem no one foresaw. It is worth rehearsing the uncomfortable questions: “Will AI be assessing us?”, “Will the test results affect bonuses?” and “Are you planning redundancies?” Managers also need to be able to say honestly when there is no answer yet.

Uncertainty about jobs calls for precise words, not magic formulas

If no decision on cuts has been made, you can say so. You must not, however, promise that there will never be any cuts if there are no grounds for such a promise. If changes to headcount are being considered, pretending the subject doesn’t exist exposes employees to needless speculation and a later loss of trust. In both cases, what you say should reflect where the decisions actually stand. A manager cannot promise on the board’s behalf something that has not been approved.

An example of an honest statement: “Today we approved a pilot in two areas. No decision has been made to change the number of roles as a result of this test. We’ll update you on the results and the next steps on 15 December.” If the company is genuinely going through a restructuring, the language must be just as specific, though it will be harder. Instead of making promises, explain how decisions will be made: who will make the decision, on what basis and when people can expect to hear. This is a matter of respect for employees, not a ploy to “sell the change”.

Every employee question needs an answer or a date

After the board’s presentation, questions will come up in the chat, in small teams and in one-to-one conversations. It is worth collecting them and sorting them by type: impact on duties, data security, quality of work, skills, assessment rules and employment. If there is no answer yet, name the person responsible for providing one and set a date. The worst thing you can do is ask people to be open and then leave the uncomfortable questions unanswered.

A simple log is useful: the question, the confirmed answer, the person responsible and the date of the update. It doesn’t all have to be public, especially when a question concerns an employee’s private situation. But it should make it possible to give teams consistent information. Managers can let the board know which issues keep coming up and which phrases cause the most misunderstanding. That information is as valuable as the number of employees who opened the announcement.

After a week, check understanding – not just whether people saw the presentation

Communication about a rollout does not end when the message is sent. A week later, it is worth asking a few teams what will change in their work, which limits on AI use they remember and what they still don’t understand. The idea is not to test people, but to check whether they know what to expect. If each team understands the rules on using data differently, what is needed is clarification, not another promotional video.

It is also good to come back to the promises of support. Has anyone been given time to learn? Is there now a procedure for reporting errors? Can team leaders say how the pilot will be evaluated? Employees don’t expect every question to have an answer already. What they do expect is that the company distinguishes plans from facts, remembers its commitments and can correct a message when it turns out to be unclear. That is how trust in the process is built, not just trust in the technology itself.

The board will not reassure people with enthusiasm for AI alone. It has to show the concrete impact on work, prepare managers, be honest about staffing issues and keep coming back to people’s questions. The credibility of the change depends more on what happens afterwards than on an impressive presentation. Employees should know not only what the company is announcing, but also when and from whom they will hear about the next decisions.

How Empatyzer helps you prepare the board’s AI announcement

Before announcing the rollout, a board member can set out the plan for Em: the scope of the pilot, the expected benefits, the decisions already made and the difficult issues that remain open. Em will help turn a presentation built on metrics into a message that makes sense to the people who will be doing the changed work. With Em, the board member can rehearse responses to questions about jobs, how output will be checked and data protection, and also prepare managers for conversations in smaller teams. The organisational analysis in Empatyzer shows the gaps between the preferred and the experienced way of working, as well as the spread of needs and motivators that may affect how the change is received. This information helps to frame questions and adjust the tone, but it is no substitute for establishing the facts about staffing. After the meeting, Em can help organise the answers, prepare difficult feedback and plan how to return to issues that still need clarifying. As a result, the AI rollout is not reduced to a one-off announcement from the CEO, but becomes a series of conversations in which employees know what is changing, what is not yet known and where they can raise their concerns.