AI agents in organizations: collaborators or threats?

“Czy sztuczna inteligencja zabierze mi pracę?” To pytanie pobrzmiewa dziś w wielu głowach – od hali produkcyjnej, przez biurowce korporacji, po gabinety zarządów. Wizja AI zastępującej ludzi elektryzuje media i pracowników. Nagłówki straszą „inteligentnymi systemami zabierającymi etaty”, a popkultura dokłada swoje (kto nie pamięta Terminatora?). Nic dziwnego, że gdy w firmie pojawia się pomysł wdrożenia AI, wielu pracowników odczuwa niepokój.

Najnowsze badanie EY potwierdza, że aż 71% pracowników obawia się powszechnego wdrażania innowacji opartych na AI – wskazując m.in. ryzyko prawne, etyczne oraz możliwość utraty pracy jako główne źródła lęku. Paradoksalnie jednak, ci sami badani dostrzegają też pozytywy – 81% uważa, że AI może uczynić ich bardziej produktywnymi. Strach ma więc wielkie oczy, ale pokazuje trochę prawdy: technologia zmienia sposób naszej pracy. Pytanie, jak się na tę zmianę przygotować – zwłaszcza jeśli mówimy o agentach AI, czyli dość nowym zjawisku w świecie biznesu.

What are AI agents – is this the beginning of a new stage in software development?

Before we move on to AI agents, let’s pause for a moment to consider the concerns themselves. Process automation and artificial intelligence have been evoking mixed feelings for years. On the one hand, there is excitement about efficiency, and on the other, fear for the future of many professions. Typical concerns of employees in large organizations related to AI include:

• Loss of job or role in the team: Will a machine replace me? The prospect of job cuts in favor of algorithms keeps many people awake at night.
• Loss of control and errors: Will AI make a costly mistake? When we hand over some of our decisions to a system, we fear errors that humans will have no control over.
• Lack of competence and understanding: Will I be able to keep up with the technology? Employees fear that they will not be able to use new tools, that they will not understand how they work, and that they will ultimately become less valuable to the company.
• Changing existing habits: “Why change something when we’ve always worked this way?” Automation often means changing established procedures, and that can be stressful—even if the change is for the better.

These fears are equally present in Poland. We have experience with automation (such as the robotization of production in factories), but the idea of autonomous AI in the office still seems like something out of science fiction to many people. However, companies are increasingly turning to digital solutions and technologies that are changing the way we work and require adaptation on the part of employees.

First and foremost, it is important to remember that automation usually applies to specific tasks, not entire jobs. However, even if technology does not ultimately take away our jobs as such, but relieves us of tedious tasks, uncertainty remains. That is why it is so important to ensure open communication, education, and small successes when introducing new solutions, to show people that AI can be a support, not an enemy.

An AI agent, on the other hand, has a certain degree of autonomy. It can be compared to a digital colleague who is given a goal and then chooses the methods and tools to achieve it. Importantly, the agent can operate in the background, initiate tasks on its own initiative (in accordance with the permissions it has been granted) and coordinate the work of other programs or agents. For example, an AI agent can analyze data, make decisions, generate reports, and even send notifications or requests to other systems—all without any manual intervention. This independence distinguishes agents from traditional scripts and bots.

Does this mean that an AI agent is Artificial General Intelligence that thinks like a human? Absolutely not. It is still a specialized programming tool, but one that is more autonomous and “resourceful” in how it performs tasks. An agent can use AI modules (e.g., language models, image recognition systems, etc.), but it will not magically solve all of a company’s problems. However, it will relieve people in specific areas.

It is also important that humans still set the rules of the game: we decide what tasks to delegate to the agent, where the limits of autonomy lie, what the criteria for success are, and when the agent should stop or ask for approval of a decision. In other words, a well-implemented AI agent is still a tool in the hands of people, not an “artificial boss” taking control of the company.

If AI agents can do so much, what does the future hold for us, humans, in an organization that uses such technology? We believe that no one will be “turned off.” The history of technological progress teaches us that certain jobs disappear, but new ones emerge—and the same is true for AI. In practice, “agentization” (if you can use that word) means the evolution of roles, not their elimination. Here’s what’s already on the horizon:

• From performer to supervisor/coordinator: When an agent takes over repetitive, routine tasks, humans can step into the role of quality controller, strategist, and coordinator. Instead of manually entering data into a report, a finance specialist will check the AI-generated report, interpret the results, and suggest business decisions. Instead of collecting data from sensors themselves, production workers will monitor a “smart” production line and respond to alerts generated by agents.
• New professions and skills: There is a growing demand for roles such as “AI trainer” or “AI analyst” – people who teach agent systems and improve their performance. For example, in customer service, consultants can co-create knowledge bases and scenarios for agents and continuously improve them. Feedback plays a key role in this process, allowing AI agents to improve their performance based on interactions with users. IT departments, on the other hand, need people who understand agent architecture so they can integrate them into existing systems and ensure that agents operate ethically and legally (this is where the role of an AI governance expert comes in).
• More time for creativity and relationships: Machines cannot replace human creativity, empathy, or strategic thinking. When agents relieve the team of routine tasks, people gain space to focus on creating new value. Salespeople can devote more attention to building relationships with key customers (while AI segments and serves smaller customers). Marketers can come up with more original campaigns (while AI independently selects the optimal budget and channels). Engineers can design product improvements instead of looking for a needle in a haystack of logs (because the AI agent will perform preliminary diagnostics). In short, the role of humans is shifting to areas where human judgment, creativity, ethics, and leadership are needed. This enables real time savings and improved team performance.

It is also worth mentioning the psychological aspect: instead of thinking of AI as a competitor, it is increasingly referred to as a “co-worker” or digital assistant. This “colleague” may not have a sense of humor (unless we program it!), but it does not get tired of repeating boring tasks and works 24/7. This approach – treating AI as a colleague – helps to allay fears. Companies that have successfully implemented AI often emphasize that “AI performs tasks, people work on goals.” Team members learn to interact with agents, just as they once learned to work with new software or an outsourcing partner. Over time, the agent simply becomes part of the work ecosystem.

This does not mean that all fears will magically disappear. It is important to maintain a clear division of responsibilities (who is responsible for what – is it a human decision or an AI recommendation), transparency of AI operations (so that the team understands how and why the agent takes certain actions), and an open culture of questioning (so that everyone can raise concerns or suggest improvements in working with AI). When employees see that they have a say in how AI is used and that the company is investing in their development (e.g., training on how to use new tools), trust is built. Regularly gathering feedback from teams is key to the success of the project and building trust in new solutions. Without this, even the best and most useful technology will face resistance.

Implementing AI agents in a large organization is not just an IT project, but above all a human project. Team leaders and management must take care of the soft aspects of this change. How can this be done in practice to allay employees’ fears and build a climate of trust?

Before you incorporate AI into your processes, talk to your team and explain the purpose of the changes. People need to know why you are considering implementing an AI agent: what problems it is supposed to solve, what benefits it will bring to the company and to them. Emphasize that it is a tool for them, not a whip over their heads. If employees are concerned about their jobs, don’t sweep the issue under the rug – address it directly. For example: “We understand concerns that automation may reduce jobs – however, our goal is to relieve you of monotonous XYZ so that you can focus on more valuable tasks.” Be open to questions. As experts emphasize, a lack of clear communication within the company only intensifies fears about AI (including fear of job loss). It is therefore better to preempt rumors with facts. Educate, show examples, and dispel doubts.

Select a small pilot project in which the AI agent will assist and involve team representatives. Let it be, for example, one specific process in the department that complains the most about excessive manual work. Invite willing employees to collaborate on the implementation – their perspective will protect you from mistakes, and at the same time, these people will become ambassadors of change among their colleagues. When they see the results with their own eyes, it will be easier to convince others. Such participation increases acceptance – people feel that they are implementing something, rather than being implemented.

Even the most intelligent agent won’t help if users don’t know how to work with it. Plan practical training for those who will be using the solution. Show them step by step how the agent works, where to check its recommendations, and how to submit comments. Q&A sessions are a good idea, during which the team can ask any questions (without judgment). You can also designate “super users” – employees who are more familiar with the technology and can help their colleagues on a daily basis. The goal is that no one feels left alone with a “black box.” The greater the competence and familiarity with the tool, the less fear and… the greater the chance that people will like their new “colleague.”

It is important to define from the outset what the agent does and does not do, and what the human-AI interaction looks like. For example, an agent can prepare a report, but the decision is made by the manager; an agent can suggest a response to a customer, but it is sent by an employee after verification; an agent can place an order for goods to the warehouse up to a certain amount, above which the approval of the purchasing department is required. This type of framework gives employees a sense of control. They know that the human role is still crucial and that AI operates within established rules. As implementation progresses, these boundaries can be gradually expanded, always communicating this to the team.

Nothing convinces skeptics like hard facts and… personal experience. Collect data from the very beginning: how much time we saved thanks to the agent, how the number of errors decreased, how the indicators improved (e.g., customer response time). Inform the team about the results, praise the initiators of the change and everyone who contributed. If, for example, an AI agent allowed you to close the financial month two days earlier or relieved the helpdesk of hundreds of trivial requests, announce it and thank people for their cooperation. Seeing real benefits, employees will be much more willing to accept further expansion of AI applications. The nostalgia for the old methods will quickly disappear when they understand how much the new tool makes their lives easier. What’s more, by publicly acknowledging the team’s efforts, you create a positive atmosphere around the change.

By implementing the above steps, you show your team that technology is not a fad imposed by management, but a well-thought-out improvement in which people are more important than the tool. Companies that take this approach to transformation gain not only in efficiency but also in employee engagement—when the team feels like they are part of the process, not victims of it.

Finally, here is a brief summary in the form of a short checklist. If you are a leader planning to introduce AI agents into your organization, make sure that:

• You have a clear business goal for implementing AI – you know what task/area the agent is supposed to improve and how you will measure success.
• You communicate openly with your team – you have explained the reasons for the changes, presented the benefits, and addressed concerns (especially about jobs) directly, honestly, and with empathy.
• You involve employees in the process – you have selected a pilot project that includes key team members, you listen to their feedback, and you give them a sense of co-creation.
• You have planned training and support – you provide training on how to use new tools, you have experts (internal or external) available to help, and you build AI competencies within the company.
• You have established rules for the agent – you have clearly defined which decisions the agent makes on its own, where human approval is needed, who monitors the AI’s performance, and how to report any errors or incidents.
• You monitor and celebrate results – you collect performance data, regularly update the team on progress, recognize people’s contributions, and adjust course when something isn’t working as it should.
• You ensure continuity of dialogue – you maintain an open discussion about experiences with the AI agent, encourage the sharing of ideas for further improvements, and respond to concerns as they arise.

We hope that after reading this article, the answer is clear: it depends on us. A well-designed and implemented AI agent can become a valuable member of the team – relieving people of monotonous work, speeding up processes, and giving the company a competitive advantage. However, the key is to lead the transformation in such a way that people feel like they are the subject, not the object.

Technology is technology, but it is trust, communication, and support that determine whether AI will be perceived as a threatening intruder or a welcome “new colleague” in an organization. Digital transformation leaders must therefore be not only innovation strategists, but also guides for their teams on the path to change. Ultimately, it is people—with the help of their digital agents—who drive the success of any organization.

EY, “Lęk przed AI. Większość pracowników obawia się rozwoju tej technologii” / “Fear of AI. Most employees are concerned about the development of this technology” (2024)

https://www.ey.com/pl_pl/newsroom/2024/01/ey_ai_anxiety

Instead of implementing technology “from above,” engage your team. Instead of risking chaos, see how an AI agent works in practice.

We’ll help you plan a pilot, prepare internal communications, and design a solution that truly supports people in their work.