Bots (short for robots) that perform automatic tasks, repeating the operations performed by a user, are widely used. Chatbots, i.e. programs with a "conversational" User Experience, are now becoming the new reality in technologies for "man-machine" interaction. To make a chatbot able to understand, respond to requests and propose solutions using everyday natural language, it is necessary to introduce very innovative context management and artificial intelligence technologies.

The peculiar elements of a Conversational AI Platform (CAP) are:

• human-machine interface based on textual or vocal conversation in natural language

• Artificial Intelligence techniques to recognize users' intentions and recommend solutions

• systematic and pervasive use of the context (position in mobility, devices used, history of previous interactions, sentiment)

• proactive system, able to send information when appropriate and not only on request

• integration with services and data.

The objectives of Research and Development are aimed at two potential uses.

Customer Service/Support: a new service model also for "customer service" (knowledge base, FAQs, questions and orders) where the partial replacement of human interaction with a self-service mode allows to scale or improve performance indicators (costs, time, availability). A chatbot has been introduced, able to respond on a static or dynamic knowledge base, useful for customer service operators.

Citizen Digital Assistant: in a broader and more challenging vision, a Citizen Conversational AI Platform is proposed for a more effective and enjoyable access to digital PA services (e.g. smart city), especially when associated with new generation channels such as Instant Messaging and voice. A constellation of voice or text chatbot services that respond to citizens upon request for information or execution of processes is under development.

2019

Also based on the AI, the search for conversational solutions for citizen services continued in 2019.

Three prototypes were produced with the goal of: 1) engineering and automating the production of bots, to quickly implement new services; 2) improve effectiveness by inserting conversation patterns, e.g. fluid conversation mechanisms, with current speech memory and list references.

Prototypes made:

• Search for general practitioners and pediatricians of the Piedmont Region on Alexa and Google Home, with refinement of the search.
  Example:
  - What time does Dr Bianchi receive?
  + I have several Dr Bianchi in my address book, in which municipality does he practice?
  - Turin
  + Which of these three? Mario Bianchi, Luca Bianchi, Giorgio Bianchi
  - The third
  + Giorgio Bianchi receives today at 4pm and the day after tomorrow at 9am.
• Search in the company directory, with direct, reverse and multi-parameter search.
  Example:
  - who deals with graphics in the company?
  - who works in the T11 office and works in the Big Data area?
  - who answers the number 321 445566?
• Search for municipal offices

2020

In 2020, we worked to improve and enrich conversational solutions, making the conversation more fluid and adding new features.

What we achieved:

• webchat support for website, Telegram, Google home and Alexa

• navigation of any relational schema (or graph)

• search of any entity (table) by attributes even of linked tables

• support for "mentions": given a list you can refer to an element by position

• the number of retrieved results " drives" the response with the possibility of further refining

• possibility of authentication via SPID

• possibility to save the favorite doctor