Enterprise Spa

Enterprise: Generative AI Drives Innovation

The use of artificial intelligence offers a strategic and innovative boost in the product platforms of Enterprise, which has already identified use cases for GenAI and started experiments in the field of artificial intelligence, also together with vendor partners

Our new product platforms (Pr.EMIA. Evolution and Core Banking Plat@one) were designed and developed to be provided to our customers also in the cloud and not just on-premises. Thanks to our certifications, also held in the application and technological service management perimeter, the hosting on which to host the platforms can also be the Enterprise cloud, offering in this case the services in SaaS mode.

The integration of GenAI

In addition to technological evolution, in terms of portability and scalability of our solutions, we are introducing, within the same platforms, some innovative components of generative Artificial Intelligence. To cite a few examples, we are starting the experimentation of a conversational user interface which, via natural language, allows us to derive structured information from our product database (for example, in the form of reports, communications to customers, etc).

This form of interaction, immediate and widely accessible, allows us to act with flexibility and effectiveness in the search for information in a broad sense, a perimeter increasingly paid attention to by our customers, for reasons ranging from monitoring to business, up to monitoring of security and compliance.

Furthermore, the use of AI allows to overcome the approach focused on the implementation of new search features, developed with the use of “traditional” programming languages, and does not require the possession of particular technical-functional skills by the user, who merely formulates a “correct” question, to obtain a result that conforms to expectations.

Two use cases

The first two use cases we have identified in this area are

1) the querying of our back-office platforms for the extraction of information requested by customers, in the branch or via a service desk: for example, retrieving the data of a generic transaction settled on one of the different possible accounts on which the applicant has any form of ownership, using any data, even partial, that refers to the operation being researched;

2) the direct query by the end customer of our digital channel platforms (retail and corporate web portals), in the same way and for the same purposes as described in the previous use case.

Trials underway for the Faqbot

The other AI initiative being tested is linked to the training of a bot that acts on the knowledge base of our solutions, to arrive at releasing a Faqbot that replaces, in a much more efficient and result-oriented way, the consultation of an online help.

In this project perimeter, we identified two other pilot use cases:

1) a Faqbot acting on the user manual of the counter application, for use by branch banking operators;

2) a Faqbot acting on the web portal user manual, for use by the bank's end customers (retail or corporate).

Vendor partners to innovate strategically

In addition to the innovations just mentioned, which we develop directly at home, we have taken on the strategic choice of integrating third-party solutions (our vendor partners) that develop less generalist AI models and more focused on certain operating perimeters.

In this context, we are concluding and will release in the coming months, the integration of our Trade Finance back-end (called TED, or Trade Evolution Dashboard) with an AI platform capable of carrying out cross-checks on the consistency and compliance of the information contained in the trade documents, against the contractual baseline defined by a letter of credit, for an international import or export operation.

In this case, our TED suite, which holds documents and contractual terms in the digital trade dossier, takes care of feeding our partner's platform, providing:

1) the Swift messages related to import/export documentary credits;

2) the digitized trade documents received from the client or counterparty.

Once the pool of information on which to activate the analysis has been received, the AI platform operates on several levels:

• congruence checks of the information reported on the documents in accordance with the contractual terms (defined in Swift messages);
• verification of compliance with the international regulatory framework that governs this type of operation (i.e., UCP 600 - Uniform Customs and Practice for Documentary, published by the International Chamber of Commerce, and ISBP - International Standard Banking Practice);
• checks on the international AML compliance of the operation (i.e. blacklists, embargo, dual-use goods lists), through a screening of the subjects, legal entities, ports and goods subject to the underlying commercial operation.

Any discrepancies and non-conformities detected by the AI engine are returned to the TED suite, which captures them in the user functions responsible for decision-making management of these potential exception situations.

Stefano Trinci,
Head of Core Business of Enterprise SpA