Business need
Managing a complex network of customer interactions
Italiaonline handles a high volume of customer interactions across multiple channels, with needs that vary widely. The company must provide timely, effective service in every interaction, whether that means giving a relevant response or creating a tailored offer based on the customer’s request.
The underlying goal is to keep improving customer engagement and the overall customer experience, from the first interaction through to post-sales support.
To achieve this, Italiaonline began exploring what AI could do for the business. They launched a broader program to adopt GenAI across their application ecosystem. Planned as a phased initiative, the program reflects the company’s commitment to exploring new opportunities to improve and innovate.
The first phase focused on post-sales operations. In this area, customer emails sets off a process that creates a case and routes it to an internal back-office team. The request is then analyzed, and the case is assigned to the team responsible for that specific issue.
Italiaonline saw an opportunity to improve this process with AI. The objective was to reduce the time service agents spend resolving each request while creating the foundation for a broader AI adoption strategy.
When selecting partners to help us define the roadmap for implementing these solutions, our existing relationship with NTT DATA and Salesforce, who were already supporting our customer service operations from both a technology and business perspective, played a significant role. It was therefore a natural choice to entrust them with guiding us on this new journey as well.”
Solution
Bringing AI into post-sales operations
NTT DATA had long supported the evolution and integration of Italiaonline’s information systems, and Italiaonline was already using several Salesforce platform components. Salesforce also bought an extensive knowledge base to support the adoption of new technologies.
The first phase of the program focused on the post-sales function, with a case-classification project that also served as the foundation for a broader GenAI adoption strategy.
When an email reaches Italiaonline’s contact center with a customer inquiry, whether the customer is requesting information, updating data, changing a product or something else, it triggers a process that creates a case and routes it to an internal back-office team. The request is then analyzed, and the case is assigned to the team responsible for that specific issue.
“We introduced an AI engine into this process through a dedicated Salesforce module,” explains Fulvio Negri, Senior Strategic Account Manager at Salesforce. “By analyzing keywords, context and the customer’s history, the system gradually learned to read and classify emails independently, without human intervention, routing each case directly to the appropriate specialist back-office team.”
The project went fully live and integrated smoothly into Italiaonline’s post-sales processes. Its success encouraged the company to move ahead with the next phase: a copilot project that provides AI-assisted support to service agents handling requests across all channels, including phone, email and chat.
When a customer submits a request, the service agent can ask the AI copilot to analyze it using an extensive knowledge base and generate a suggested response. The service agent then reviews the proposed reply before sending it to the customer.
We introduced an AI engine into this process through a dedicated Salesforce module. By analyzing keywords, context and the customer’s history, the system gradually learned to read and classify emails independently, without human intervention, routing each case directly to the appropriate specialist back-office team.”
Outcomes
Greater efficiency in customer service management
Case classification has significantly reduced the manual effort involved in handling routine tasks.
“Just a few months after the project launched, around 70% of cases are correctly assigned to the appropriate operator and 14% are handled entirely by the AI,” says Mariarosaria David, IT, CRM, Marketing and Call Center Manager at Italiaonline. “These are typically requests that do not require human intervention. The system not only classifies the case but also automatically sends a response to the customer and closes the ticket, resulting in a fully automated process.”
By improving the initial triage stage, the project has also improved the overall customer experience. Average case resolution time has fallen from 2.28 days to 1.76 days.
“In one year of operation, the AI made only one significant error. It has been an important and highly successful project for Italiaonline,” concludes David.
“During testing, whenever we found differences between manual case assignment and the AI’s recommendations, we initially assumed the algorithm needed further refinement,” adds Andrea Fravega, Senior Vice President, Telco & Media at NTT DATA. “However, after reviewing those cases more closely, we discovered that the AI’s classification was actually more accurate. That further demonstrates the effectiveness of the solution.”
“The benefits of the copilot project cannot yet be measured because it is still under development,” explains David. “However, we expect it to improve both the speed and quality of responses to customer requests while also reducing operator training costs.”
In addition, generating responses through a consistently trained AI engine is expected to improve the consistency of Italiaonline’s tone across customer communications. This will reduce the variation that naturally occurs when responses are written manually by different operators, each with their own writing style and approach.
Having established a practical foundation for using AI, the next milestone on Italiaonline’s roadmap is the introduction of AI agents.
A future built on AI
NTT DATA and Salesforce have developed a proof of concept built on Salesforce Agentforce, enabling AI to interact directly with customers using natural language. In response to a customer request, the system can independently create and classify a case, complete the required action, close the case and send a response to the customer.
The proof of concept was demonstrated using information requests and customer data updates submitted through chat, although the same approach can be extended to a much wider range of use cases.
“It’s almost an evolution of the copilot because the AI no longer just supports the operator — it takes action itself,” says David. “We’ve seen an end-to-end prototype that operates entirely autonomously. That’s the direction we want to pursue in the near future.”
Italiaonline’s AI roadmap extends beyond autonomous agents. Two additional initiatives are already planned:
- The first will use AI to predict customer churn, helping the company identify customers who are at risk of leaving and proactively launch retention activities.
- The second focuses on cross-selling. By analyzing customers’ purchase history, the solution will recommend the offers they are most likely to accept. Using statistical models, the same approach could also be applied to prospective customers.
These initiatives are already in the planning stage. NTT DATA, together with Salesforce, has supported Italiaonline in defining the business context and strategic direction. The projects will now move into detailed design before implementation begins.
“The path ahead is clear. For Italiaonline, this is a journey made up of multiple stages, each addressing increasingly complex challenges through a phased, multidomain AI adoption strategy,” concludes Fravega.