Case Study

AI for knowledge management: the customer service engine

An intelligent platform that transforms customer service, improving access to information and user satisfaction with faster and more accurate answers.

#Customer Service
#Knowledge Management
#AI & Azure

Picture

THE CHALLENGE

Optimise quick access to information for customer service operators, overcoming fragmentation and inefficiencies.

Scenario

The challenges of outdated knowledge management

With more than 3,000 customer service operators handling thousands of telephone requests every day, one of the largest Italian service operators was slowed down by an obsolete knowledge management system. The operators found themselves navigating through duplicate, poorly classified and dispersed information on different platforms. All this made simple operations, such as retrieving details of a file or answering a frequently asked question, much longer and more complicated than necessary. Every minute lost in these searches not only slowed down the service, but also compromised operational efficiency and the overall experience of end consumers. With the growing need to provide quick and accurate answers in a short time, the customer was looking for a solution that would optimise the way in which information was managed, and improving the quality of the service in every interaction.

Picture

The solution

Generative AI and Azure to optimise customer service

To break down inefficiencies and optimise customer service, Cluster Reply has developed a knowledge management platform powered by AI that has transformed access to information. Thanks to the integration between GPT-4 and Microsoft Azure services, operators can consult the knowledge base by speaking to the system in natural language, obtaining immediate answers at the same time as the telephone conversation with the user.

The system combines Azure Cognitive Search with a Retrieval-Augmented Generation (RAG) architecture to ensure that information is always fast and accurate, eliminating fragmentation and duplication of content that affected performance. Supported by Azure Cosmos DB, the platform manages high volumes of data while maintaining high efficiency and offering essential scalability for the needs of a large national operator.

HOW WE DID IT

From fragmentation to centralisation: towards efficiency

We started the project by centralising all content and documents in a single repository, eliminating duplications and rationalising access to information. Using Microsoft Azure, we have configured the infrastructure components necessary to organise and make documentation accessible and enable advanced search services. In this way, we guaranteed that the system was able to handle large volumes of data and to adapt quickly to the need for quick and accurate responses from customer service operators.

The platform was gradually activated starting from a pilot group of 300 operators, and then extended to all customer service operating units. During the initial phase, we used stress testing techniques to validate the quality of the responses generated by the system and optimise the prompting based on the operators' most frequently asked questions. At the same time, we conducted targeted training sessions for operators, providing practical tools to make the most of the AI platform and interact with the knowledge management system efficiently. Finally, we have established a continuous monitoring system, to detect any anomalies and constantly improve the performance of the platform, gradually adapting it to the evolving needs of the customer.

The results

Satisfied operators, reduced errors

The new AI-based knowledge management platform led to a significant improvement in operational efficiency and the quality of customer service, with measurable impacts at all levels of the organisation. Among the most significant results:

Picture

50% reduction in average response times to customer requests.

Picture

Successful implementation on a base of more than 3,000 operators, distributed across multiple operating units.

Picture

Centralised management of more than 7,000 documents, completely eliminating duplication.

Picture

Operator satisfaction rate at 91% out of more than 30,000 requests handled.

Picture

Greater accuracy in information retrieval, with a 40% reduction in errors.

Picture

Cluster Reply is a Reply Group company specialized in consulting and system integration services based on Microsoft technologies. With in-depth knowledge of the different market segments and thanks to Microsoft certifications, in the Modern Work, Business Application, Digital & Apps Innovation, Cloud & Infrastructure, Data & AI, IoT & Security, CRM, Low-Code and Sustainability areas, Cluster Reply creates solutions that guarantee efficiency, flexibility and security.