A virtual assistant capable of continuous evolution

The Jiminy project (name inspired by Jiminy Cricket, Pinocchio’s “Talking Cricket”) aims to represent a breakthrough in the adoption of artificial intelligence by small and medium-sized enterprises (SMEs). The goal is clear: to make the potential of AI accessible also to those companies that, due to their limited resources, struggle to keep up with large competitors.

With the Jiminy project – co-funded by ARTES 4.0 – a new avenue is opened up: the development of a intelligent business communication system that automatically adapts to the needs of the enterprise. At the heart of the project is a Enterprise Virtual Assistant (EVA) that learns directly from day-to-day communications between customers and the company, without the need for specific training by staff. Every phone call becomes a valuable source of data, from which the system automatically draws information to improve customer service and optimise workflows, without interrupting normal activities.

A talking cricket for the democratisation of artificial intelligence in customer service.

In a context where digital transition is crucial to remain competitive, Jiminy offers a concrete opportunity to bridge the gap with large companies, enabling SMEs to exploit AI in a simple, immediate and accessible way: Despite the significant benefits that SMEs could gain from exploiting the potential of new Artificial Intelligence technologies, they remain significantly behind larger competitors.

Indeed, while 61% of large companies have already initiated at least one Artificial Intelligence project, only 18% of SMEs make use of it: a gap that, without the adoption of targeted solutions, risks widening further. Jiminy was created precisely to bridge this gap, offering SMEs an affordable system that is easy to implement and able to automate customer care processes in real time, without the need for in-house technical resources.

VUI (Voice User Interfaces), CUI (Conversational User Interface), Intelligenza Artificiale e Machine Learning

VUIs (Voice User Interfaces) represent a significant evolution from traditional GUIs (Graphical User Interfaces): whereas GUIs are based on visual interactions through icons, buttons and menus, VUIs allow interaction through voice commands, providing a more natural and intuitive experience for users.

A key aspect of VUI is the importance of artificial intelligence (AI) and machine learning. These technologies enable systems to understand natural language, improve speech recognition accuracy and progressively adapt to user preferences. AI enables VUIs to handle ambiguity and contextualise requests, while machine learning helps to continuously improve performance through the analysis of usage data.

The leap from GUI to VUI, however, involves a compulsory step through the development of the CUI (Conversational User Interface), i.e. an interface that allows the user to interact with a computer system through natural language, simulating a conversation similar to that between human beings.

Real-time customer suggestions and support

Kalliope Jiminy is a EVA capable of creating its own Knowledge Base autonomously, bypassing the need of other EVAs currently on the market to have a large knowledge base and resources to carry out specific manual training in order to function.

La funzione aziendale target di Jiminy è quella del “Customer Care” (CC). This function was chosen because it cuts across almost all companies and is usually characterised by major scalability problems due to the great stress that staff, under the high pressures typical of the sector, experience. In addition to an EVA capable of supporting CC operators during normal work activities, Jiminy aims to create a co-pilot capable of facilitating the training of new resources as well, since it will be able to automatically generate training calls, simulating conversations based on past interactions, with which new operators will be able to experiment and learn how to handle similar situations in a context of comfort and safety.

How Jiminy works

When a customer calls, he is greeted by a voicebot that performs an initial triage of the problem; the information gathered is used to automatically create a ticket, which is then verified by a human operator. During the call, Jiminy listens and analyses the conversation, providing real-time suggestions to the operator based on its accumulated knowledge. At a later stage, this same knowledge is reused to create simulated scenarios with which new recruits can “train”.

The various steps:

1. Call reception: A voicebot greets the customer, collects the customer code and performs an initial classification of the problem. The call is routed to an operator.
2. Creation of knowledge database: the EVA “listens” to the exchange between customer and operator and a db is created (according to GDPR standards), with the aforementioned records;
3. Ticket opening: Information extracted from the telephone interaction is reprocessed and used to automatically generate a ticket, which is then verified by an operator.
4. Real-time assistance: During the call, Jiminy suggests solutions to the operator in real time, via a conversational interface, using knowledge accumulated from previous interactions.
5. Continuous learning: Each call is recorded and analysed semantically, contributing to the enrichment of the knowledge base.
6. Staff training: re-use of company know-how to generate call simulations based on real cases, allowing new operators to practise in a safe and controlled environment.