More projects, especially because of the health crisis. But also a declining level of satisfaction … AI and customer relations are still looking for each other.

“In the field of customer relations, the scope of possibilities of AI is vast. Many banks and insurance companies are already using natural language AI for automated processing of emails from their customers or for managing claims through automated verification of attached documents. They were able to achieve levels of precision and response rates higher than what human treatment could provide “, observes Eric Frances, VP Sales France & BeNeLux, expert ai.

Of course, the health crisis has accelerated projects. Many companies have rushed to deploy online services. Often, moreover, to respond to a specific problem. And so, unfortunately, without taking into account the overall strategic potential for the organization.

“Artificial intelligence has come out of the stage of experimentation and fantasy,” said Eric Frances. And that’s just the beginning ! According to Capgemini, in just two years, the daily use of AI-based services in interactions with businesses has increased from 25% in 2018 to 40% in 2020. But if the crisis and the associated physical distancing have not fact that amplifies the phenomenon, the level of user satisfaction has fallen by 16 points since 2018. Expectations are high. Too much ?

What customers want

To meet a customer’s demand, you must first understand it. It sounds trivial, but from a technological point of view, understanding the language – and especially its approximations – remains a real challenge, says Eric Frances. “This is even the main issue for automated tools for interacting with customers, such as chatbots, messaging systems and remote support. Today, it is not uncommon for a customer to be forced to rephrase their request multiple times to match company vocabulary and be understood by the tool. This is obviously detrimental to the quality of the customer experience. However, if a company decides to use AI for part of its customer relationship, it must allow it to offer its customers a better quality of service as a whole. “

Understanding natural language: a mission for a specialized AI

The purpose of artificial intelligence applied to the understanding of natural language is precisely the analysis and extraction of information from so-called “unstructured” textual content, that is to say that it is about understand the context, concepts, expressions, key messages, data (values, dates, etc.), tone, feelings etc. without ambiguities, and to make all this collected information usable in an automated way. AI solutions applied to understanding natural language rely on machine learning or semantic rules built from a knowledge graph.

According to IDC estimates, unstructured data will represent 80% of global data by 2025. This data contains operational and strategic information that is currently largely under-exploited. In question, the specialization of the suppliers of AI technologies in a precise field of application: chatbot, telephone assistance, text analysis in social networks… “This compartmentalized approach does not make it possible to capitalize on the development of the business lexicon and on machine learning algorithms for application to other projects. In this context, as soon as we change the mission of an AI, it must start from scratch. However, AI applied to natural language understanding can and should be part of a holistic approach and be leveraged in multiple use cases within a single information system. “

Make AI your new customer advisor

Yet the potential is there. In some cases, the automation of the process by AI can go as far as triggering compensation for amounts not exceeding a few thousand euros, and thus reduce processing times from several hours to just a few minutes. . It also frees up time for employees who can focus on the most complex files, helping to improve service and the customer experience.

Technically, tools based on AI can be very satisfying. But for them to really participate in the performance of the company, they must be integrated into a global strategy involving all the businesses concerned, notes Pierre Frances. “The best approach therefore remains that which combines artificial and human intelligence, in order to best meet consumer expectations and deal with unforeseen situations. At the start of the pandemic, for example, some chatbots were unprepared for questions about COVID-19. The best-equipped companies were nevertheless able to react quickly and adapt their services in barely two weeks, which represents a major advantage in terms of customer satisfaction, and therefore competitiveness. ”