AI agents mean a breakthrough in the digitization of banks

Definitely yes. Both banks and insurance companies are at the forefront of companies that implement AI solutions. Research and development are particularly important to them. The report, which was recently published by Boston Consulting Group, “For Banks, The Ai Reckoning is here”, shows that every fourth company from the financial sector has implemented such solutions. The essence is to integrate artificial intelligence with key processes. Companies are looking for implementations that will be effective. They expect that savings and new sources of market advantages will appear following investments in this area.

Our experience could be said, according to the fact that we are investigating the limits of what is possible in the field of broadly understood artificial intelligence. And in fact, we see a great future in developing AI agents. Artificial intelligence ceases to be an addition or curiosity, but becomes the foundation of modern banking.

This is a very significant quality change. Chatbots are reactive tools, and thus answering questions and solving simple tasks. AI agencies are autonomous systems. They can analyze data from many sources, independent of each other. They can make decisions and act proactively. An example is the ability to detect dirty money by agent, initiating the verification procedure and notify the relevant departments in the financial institution. This is a completely new level of operational intelligence automation. To sum up, AI agents, first, act autonomously, and secondly, they integrate data, thirdly, learn and adapt to changing conditions and needs. Fourthly, they work with people.

No, though with some reservations. Artificial intelligence actually takes over routine tasks, such as answers to customer inquiries or document analysis, but this does not mean mass layoffs. BCG research shows that 70 percent The institutions invest in the development of employees’ competence in the context of cooperation and efficient use of AI. The role of man at work changes from the contractor to process designer, analyst or strategist. AI is a source of evolution, not revolutionary changes in the labor market.

This is a reasonable question and one of the key challenges facing us. AI agents work on high sensitivity data – personal data of clients or transaction stories or internal compliance policies. Hence, safety and compliance with regulations must be an integral part of the architecture of these solutions. At BCG Platinium, we recommend the Secure by Design approach. From the very beginning, it assumes the design of AI agents with security in mind. An element of this approach may be, for example, local implementation of large language models or the use of Retrieval-Augmented Generation architecture. It is about the approach combining the capabilities of large language models (LLM) with access to external sources of knowledge, such as documents databases, search systems or data sets. Instead of using public LLM models, more and more entities are implementing local entities, e.g. in a private cloud or on their own infrastructure.

This ensures full control over customer data and meets the regulatory requirements. I mean GDPR or KNF guidelines. In addition, these solutions allow you to adapt architecture to the specifics of the organization and its needs. This guarantees the agent’s operation in a safe and lawful manner.

AI agents can carry out tasks that are associated with collecting a large amount of data. These can be both internal tasks and those requiring contact with the client. An example would be the automation of activities related to credit risk analysis. The agent can analyze credit documents sent by the client and issue a recommendation that will be approved by the analyst. AI agents will also play a key role in intelligent debt collection, where systems will autonomously negotiate repayment plans, propose restructuring or managed the case in accordance with the debtor’s profile. AI agents can finally test software or automate the production process, i.e. implement Devops/Gitops.

We are at such a stage when we recommend our clients so that it is a man -supervised environment. A bank employee receives a suggestion from the AI ​​agent and makes a binding decision. So it is about the evolution of the role of man, from the contractor to the supervisor of the processes implemented by artificial intelligence.

We have great potential, in Poland there are many good technical universities educating talented engineers dealing with the development of artificial intelligence. The startup ecosystem is also growing. However, in terms of the implementation of AI agents on an industrial scale, we are behind. In Western Europe, over 30 percent companies have already implemented AI at the strategic level, in Poland about 12 percent We need more courage and a tendency to experiment. Many companies still approach innovation too conservatively. Meanwhile, technologies such as broadly understood artificial intelligence is the future of business and an indispensable source of advantage. You also need better cooperation between the private and public sectors.