How Agentic AI Is Transforming Access to Digital Financial Services

Once upon a time, there were APIs; today, there is Agentic AI.

"For years, software platforms spoke the same language-that of APIs," says Daniele Savarè, Chief Technology Officer at Finomnia. "The digital model was based on a simple principle: all systems exposed functions through endpoints, with a rigid connection between caller and callee, and every integration required changes to the API catalog. AI, however, has shifted the focus, and today it is no longer endpoints that define what can be done, but AI agents that orchestrate services to achieve the goal."

AI is a cross-functional tool

Whereas platforms were previously built as sets of separate modules, and their true value lay precisely in the ability to assemble these components, with Agentic AI, platforms are transformed into intelligent environments populated by autonomous agents that reason, query, compose, and act.

"Before generative AI, integration plans could take a year or longer because we had to integrate processes, design interfaces, and integrate the back office," notes Savarè. "AI has changed the perspective, leading us to rethink how platforms function, communicate, and are used. We are building an Intelligent Platform that does not think in terms of vertical products, but rather in terms of capabilities accessible to AI across our core systems (Banking, thanks to the integration of Panda into core banking solutions; Lending, with the merger of OCS and Qinetic into the Finomnia group; Leasing and Factoring, with the expertise of Forward 3000 and K4F, respectively, Ed.)".

Features are no longer organized into blocks

Every part of the platform becomes potentially accessible to any agent through unified protocols, without the need to design predefined processes or integrate vertical channels: "The agent understands the intent and uses the capabilities to build a personalized flow that adheres to policies and is perfectly governed," continues Savarè. "The functionalities are not software blocks tied to a module, but intelligent tools exposed to a common agent-based layer, governed by rules, security, and shared semantics."

Security is built-in

The shift to an ecosystem populated by agents actually amplifies the need for oversight, especially since regulations such as the AI Act and DORA require monitoring of their use to ensure security, traceability, and compliance with European requirements.

"Agents are not entities free to act on the system, but digital operators working within the platform's boundaries, using only what has been explicitly authorized," explains Savarè. Security is built-in, and policies, permissions, operational limits, and scopes are applied in accordance with the control practices already in place today. Within the Finomnia group, altermAInd brings products specifically designed for this scenario."

A new access channel is opening

These AI-centric platforms, accessible to both human users and AI agents, can serve as the first point of contact between people and financial services in the Agent-driven era. How? By opening up new sales channels for digital intermediaries.

"In the short term, the intelligent and universally integrable platform will be able to adapt to channels and contexts: designed to thrive in the world of social media, voice assistants, AI marketplaces, and personal and corporate agents," explains Savarè. With the agent-based model, technological transformation meets commercial transformation: agents query the platform, configure solutions, prepare simulations, and generate proposals and contracts. They even verify technical feasibility thanks to AI Protocols (see box), such as in the case of pre-configuring a financing or leasing proposal, built in real time based on the user's intent."

An open and interoperable ecosystem

It is a model that naturally aligns with embedded finance, enabling financial services to be integrated into apps, AI marketplaces, or personal assistants.

"Finance reaches the customer exactly where they are," comments Savarè. "The functionalities are exposed via standard agent-based protocols, so that third-party actors and agents can use them independently, eliminating complex technical integrations, and the platform becomes a shared infrastructure on which customers and partners can build and combine their financial services in an open and interoperable ecosystem."

AI Protocols: What They Are and How They Work
In the era of agent-based artificial intelligence, AI protocols are what Application Programming Interfaces (APIs) have been over the past twenty years: the foundation of interoperability. Unlike APIs, however, agent-based protocols do not describe "how to call a function" but rather "how an agent can understand, use, and combine a capability."

"The Model Context Protocol (MCP) allows agents to view tools and data in a structured way, while the Agent-to-Agent protocol enables agents to collaborate, and the Agent-to-UI protocol makes interfaces dynamic, generated or enriched by AI in real time. These protocols are the layer that makes the platform truly universal," Savarè points out. "There is no longer a need to define every process, screen, or integration in advance, since the agent interprets the rules, respects boundaries, and autonomously composes what it needs."

AziendaBanca - Written by G.C. on May 26, 2026