The financial industry is incorporating AI agents; in Argentina, implementations with cost reductions and improvements in efficiency are already known
“This is not the future: it is what is already published in annual reports and press releases,” Caramutti added and added other metrics: ING (March 2026) is already reassigning 1250 positions to AI supervision. HSBC identified 2 to 4 times more suspicious activity and reduced its alerts by 60% with a system that completes investigations on days where it used to take weeks. JPMorgan removed 95% of its AML (prevention of money laundering) false alarms.
LA NACION was in conversation with numerous banks operating in Argentina and also asked them about metrics that show the impact of the incorporation of agent AI. “75% of digital inquiries are resolved in the first contact and attendance at branches fell by about 30%, which reflects that more and more operations can be resolved simply and securely from digital channels. Today, about 90% of operations are already carried out digitally,” revealed Fernando Turri, Director of Technology, Operations, Data and AI at Santander Argentina.
In Galicia, the improvements are observed in a reduction in management times, thanks to faster access to the client’s context, an increase in resolution in the first contact and a lower need for referrals in frequent consultations. “The answers begin to reflect their specific situation, considering what they have already done before or if they have an ongoing management, which avoids having to explain everything again or look for information in different places,” explained Diego Baccini, Head of Channels in Galicia.
In addition, internally, AI is also transforming the banking industry, as it not only improves the relationship with the customer, but also gains efficiency. “In one of our main tribes, we see results such as 80% of the code generated with AI tools, 33% improvement in Median Lead Time and 50% increase in development speed,” said Rosenbolt.
Next steps
Against this backdrop, the great unknown that arises revolves around how the Agent AI will continue to transform the banking world and the rest of the industries. From N5, they anticipate a somewhat futuristic scenario for the next time: an AI that knows us so well, that offers services and products, but in real time. “Today there is no insurance for a few hours, but with the advance of the agents, it could happen that, as they know your agenda, they know that you will have to go to a dangerous area for a meeting and will offer you insurance, with the possibility of contracting it for a few hours,” Colombo exemplified. In short, they will have the versatility to create a product in seconds, with the possibility of customizing it meticulously according to the needs of each customer. “This model already exists in other countries,” Colombo explained and specified that some have the possibility of knowing your location, knowing your interest in buying a coffee and recommending in real time a nearby coffee shop, with which your bank has a discount.
That future is not far from reality, since, at MODO, its next steps are aligned precisely in that direction. In the words of Sebastián Rosenbolt, CTO of the company: “We are thinking of new experiences, where the app gets to know you better and improves the ways of recommending super specific and relevant promotions, or created especially for each user”.
Banco Nación also anticipates some upcoming functions: delegated payments, proactive assistants that anticipate needs, intelligent fraud alerts that not only notify, but also suggest actions, personalized financial advice, automatic treasury management for SMEs, and 24-hour accessible service in all branches in the country. In their case, they clarify that their strategy is not to replace traditional channels, but to add the agent channel as one more option, always maintaining the possibility of human attention.
At Santander, the bet is similar, although they add other skills that an AI agent will be able to execute: assisting in investment decisions, comparing credit alternatives, recommending savings strategies, suggesting refinancing preventively, anticipating financial imbalances and helping to organize personal financial goals. “Banking will stop being primarily transactional and will become an assisted experience. The customer will not only operate, but will also have a “financial co-pilot” available 24/7,” Turri summarized.
The future of work
Progress is rapid and increasing, but regulation for this type of progress does not yet exist. “The agent economy raises new questions about responsibility for autonomous decisions, the digital identity of agents, interoperability between institutions that will probably require specific guidelines from the BCRA, the UIF and the AAIP in the coming years,” said Banco Nación spokesmen.
On the other hand, the changes for employees seem to follow the same direction: there is talk of a transformation in roles, rather than a disappearance. “I believe that agents will never be able to replace soft skills and human interaction, that’s where profiles become more valuable and manage to make a difference,” Rosenbolt anticipated. Specifically, they are working on a new proposal, which reflects this view: “We are thinking of several groups of agents that can communicate with each other to carry out complex tasks, under a model that we call “middle-to-middle”: unlike the end-to-end model where AI tries to do everything from end to end, here people remain at the extremes – defining the problem well at the beginning and verifying the solution at the end – while agents work on the medium, coordinating with each other to execute the different parts. It is a model where humans and AI complement each other without either losing prominence where it matters most.”
In fact, they insist that customers value the improvements that AI agents have introduced, but expect to find human support when the situation requires it. “That’s why the future is not ‘AI versus people,’ but an intelligent integration between the two,” Turri explained. In this context, Baccini pointed out that many simple queries are now resolved digitally, but “the role of the human advisor becomes more relevant in complex cases, where criteria, accompaniment and the ability to resolve are needed.”
