How AI Priorities Changed in Financial Institutions in 2025?

Artificial intelligence in the finance sector was discussed under similar themes for a long time. It was mostly confined to customer service; playing a supportive but limited role in simple question-and-answer scenarios. This approach was beneficial to a certain extent but did not create lasting transformation in the field.

2025 became the year this cycle was broken. AI projects were no longer evaluated by the question “does it work?” but rather “does it really deliver business results?” For CBOT, 2025 was exactly the period when this perspective became clear. Both in terms of growth and product, targets were surpassed; active collaboration began with thirty financial institutions, including fifteen banks. However, beyond these numbers, the most significant development was the shared realization that generative AI had evolved from a theoretical promise into a practical, value-generating structure in the finance sector.

From Chatbot to Digital Workforce

Entering 2025, one fundamental priority was clear: not positioning AI merely as an interface or support tool. It was evident that AI in financial institutions had long been confined to a single domain. Yet the real need observed on the ground was for solutions that could truly ease the burden on teams working under operational pressure and integrate naturally into existing processes.

At this point, the focus moved far beyond just answering questions. Digital Employees who could understand processes, build context, and fully undertake specific tasks came into focus. The year 2025 marked the concrete implementation of this approach, and by year-end, 26,000 active Digital Employees had become operational in financial institutions.

Artificial Intelligence Positioned in the Right Place

One important principle remained unchanged throughout this transformation: don’t place AI everywhere. The goal was not “more AI,” but AI that works in the right place. For this reason, the focus was on areas where human resources were heavily concentrated, tasks were repetitive yet required attention to detail, and compliance with regulations was critical.

When it comes to the finance sector, efficiency alone is not enough. Trust, auditability, and sustainability are just as crucial. Therefore, when designing Digital Employees, the fundamental reference point was always the same: under what rules and boundaries a human would perform the same task.

Why Did Expectations Change?

In the past, expectations from AI in financial institutions were quite limited. It was often sufficient for it to work 24/7, support call centers, and automate a few routine tasks. AI was not given decision-making responsibility and was mainly positioned as a helper.

In 2025, this approach changed significantly. Institutions began to view AI not merely as an auxiliary system but as a structure that understands the process and works as an integral part of operations. This shift was made possible not only by technological advancement but also by institutions learning how to position and adopt this technology effectively.

Key Focus Areas Carried Into 2026

As we approach 2026, several areas are emerging as priorities. Debt collection on overdue receivables leads the way. In this area, a digital employee has been developed that can establish natural communication with customers, provide debt notifications, secure promises of payment, and manage the process end-to-end. This approach creates significant impact both in terms of operational efficiency and customer experience.

However, text-based solutions alone have started to fall short. Interest in voice technologies has notably increased, and solutions focusing on specific areas of expertise are seeing growing demand.

Cost Is Now On the Table

Another key change has been on the cost side. GPU and processing costs are no longer just a topic for IT teams but are now on the radar of finance teams as well. AI investments require significant budgets, and there’s now a clear demand to see measurable returns.

At this point, increasing GPU efficiency through special inference architectures has become critical. The goal is to control costs without sacrificing performance. 2025 clearly demonstrated that efficiency and profitability can advance together with the right architecture.

2025 was the year when AI in the finance sector evolved from a support tool into a system architecture that creates tangible business value within organizations. This transformation was shaped not only by technology but by how institutions embraced and integrated it.