From digital banking to a model that anticipates needs
In the financial sector, artificial intelligence is no longer considered merely a digitalization tool but a central factor of transformation. This is the thesis presented at the Cetif Summit 2026, the event promoted by the Centro Tecnologie, Innovazione e Servizi Finanziari at Università Cattolica, which brought together managers and specialists from banks and insurance companies in Milan. According to Federico Rajola, director of Cetif, Banking 3.0 marks the shift from a bank that delivers services through digital channels to a model capable of understanding the customer's context, anticipating their needs and proposing personalized solutions. An evolution made possible by the convergence of machine learning, generative AI, agentic AI, cloud computing, advanced data analytics and automation.
The main novelty, according to the analysis presented at the summit, concerns moving beyond predominantly reactive service models. AI is already changing processes ranging from risk management to distribution, and from support to advisory. However, the most dynamic institutions are already looking toward 4.0 prospects, where autonomous agents are integrated into operational processes and not only into customer support.
Chatbots, credit and fraud: where AI is already operational
In customer relationships, traditional chatbots are evolving into agents capable of interpreting complex requests and supporting complex operations. Citet data indicate that 11% of chatbots use generative AI and that 22% of financial institutions are experimenting with it. On the offer personalization front, data analysis enables services more consistent with customer behavior; generative AI is consolidating in marketing and sales and beginning to touch typically banking processes. The agentic world, by contrast, remains further behind, with the first use cases being tested in the development of marketing plans and the management of product portfolios.
In credit and risk management, machine learning supports information analysis and decision-making processes, while in fraud prevention it helps identify anomalous behavior and suspicious transactions. Cetif nevertheless stresses that the spread of technologies must be accompanied by skills development and the evolution of organizational models, especially when intelligent systems acquire greater operational autonomy.
Insurance: less bureaucracy, more consulting
In the insurance sector, 55% of agency staff time is absorbed by administrative and management activities, versus 30% dedicated to sales activities. For Cetif, this figure highlights AI's potential to free up resources to be allocated to advisory: in perspective, many routine processes could be handled by AI agents. The analysis also finds that more than 50% of national agencies consist of a single financial agent with fewer than three subagents, and that in these entities Auto products account for an average of more than 65% of the portfolio.
Generative AI tools can support intermediaries in preparing quotes, analyzing contractual documentation and identifying protection needs, making clauses and exclusions more understandable. Similar applications concern underwriting, risk assessment, claims management and predictive marketing. The goal indicated by Cetif is not to replace the intermediary but to enhance their capabilities by reducing repetitive activities.
The open challenge: technology faster than the organization
A final element concerns customers, who, thanks to conversational tools, arrive at the meeting with the advisor already having preliminary information. Institutions are called upon to offer simple and personalized digital experiences. Rajola observes, however, that technology is racing ahead while organizations, skills and governance models have yet to evolve: the real challenge will be turning AI's potential into concrete value. The actual adoption of agentic systems and the organizational impact of these changes remain to be verified in the field.
