Darko Marjanović, director of Things Solver, a member of ASEE Group - Digital workers and agentic AI are already transforming business in Serbia

He adds that these assistants help companies overcome labor shortages, optimize costs and free up employees` time for creative and strategic tasks. We discussed how agentic AI works, where it is already being used in Serbia and what changes it will bring in the coming years.

- Agent AI, or agentic AI, uses large language models not only to generate text, but also to perform tasks independently. For example, an AI sales agent for banking products can read a catalog with all relevant information on its own, launch an application for a customer and manage the entire process that a human would normally do. This enables companies to perform entrusted activities faster and more accurately, without the need for complex coding. Agent AI brings digital assistants closer to human capabilities, improves collaboration between people and systems, and opens up space for employees to dedicate their time to more complex and strategic tasks.

In contrast to traditional AI models, which respond to individual queries, agent AI functions as a standalone operating system. It breaks down complex tasks into a series of smaller steps, executes them automatically, but leaves people with the ability to monitor and correct when necessary, making digital assistants an integrated part of teams.

- In banking, AI agents are mostly used to optimize back office processes, such as document preparation, loan approval or various checks. In this way, the risk of errors is reduced and the process is accelerated, which directly affects efficiency and customer satisfaction. For younger generations, who are used to apps and chatbots, interacting with an AI agent becomes a natural part of the process, while for older users it is still a question of trust.

- The Workforce Hub platform was developed to enable companies to create AI agents with clearly defined skills and actions. Each agent has tasks it can perform, knowledge it relies on, and protections that prevent errors or unwanted actions. In this way, the agent can perform activities independently, while the management still has complete control over the process.

The platform enables companies to monitor the work of agents, evaluate results and ensure that all activities are in accordance with internal rules and legal regulations. The advantage of this approach is that small and medium-sized companies can implement digital assistants and improve business operations without large IT resources. The Workforce Hub makes it easy to integrate agentic AI into everyday processes, connecting practical application and security into a single solution.

- The most important advantage is greater operational efficiency, as the AI ​​agent takes over routine and standardized tasks, allowing employees to focus on strategic, creative and valuable activities. Additionally, agent AI reduces the possibility of errors in processes such as customer support or document processing. Digital assistants work consistently and can operate 24/7, which further improves overall efficiency. The presence of digital assistants relieves staff and allows companies to optimize costs, because sometimes one agent is sufficient for certain activities instead of new full-time employees. This is particularly important for industries facing a shortage of skilled labor, such as banking, retail or telecommunications.

- They are mostly applied in banking, telecommunications, retail, and the oil and gas sector. In these industries, there is a great need for standardized processes where digital assistants can speed up operations and improve customer support. Interest is also slowly growing in other sectors, such as hotels, human resources, employment and logistics. Any company that owns a significant amount of data or faces repetitive tasks can benefit from AI agents, as they enable faster and more accurate processing, freeing up people`s time for more complex decisions and customer interactions.

- The AI agent can operate in three modes:

In practice, this means that employees no longer spend hours checking documentation or answering basic customer questions, which reduces stress and improves overall efficiency.

- One client even gave his AI agent a name and treats him as part of the team, which shows how integrated digital assistants are already in the day-to-day work of companies. Response time to inquiries has been reduced from tens of minutes to a few seconds, and users can now quickly find information about products or services before contacting employees. The agent is available on the web platform, tablets, in-store screens and mobile devices, which enables a consistent and high-quality user experience. This approach shows how much digital assistants can improve operations and relieve people from routine tasks.

- The next step in the development of agentic AI is the additional improvement of security and data access control, as well as integration with business process modeling systems. The advice to companies is to approach the implementation of AI agents in a deliberate and systematic way. It is important to clearly define the agents` tasks, centralize their management and control, and monitor the results to avoid unintended consequences. Digital assistants are not there to replace employees, but to be a support and work partner. Their application enables companies to improve operational efficiency, reduce errors and improve service quality, while leaving employees more time for more complex and creative tasks.