Every day, the customer support experts of the rental department of TVH Equipment manually process a big amount of orders that arrive via email. Usually, these documents arrive in an unstructured manner: via text in an email or attached via PDF documents. Processing these documents manually is a repetitive task that takes up an enormous amount of time, resulting in a heavy workload with less time for customer interactions. In addition, the repetitive nature of this task and lack of control led to errors creeping in.
To process these high volumes of emails, TVH Equipment’s customer support mainly focused on retyping these orders into their ERP software. Which is crazy... Typing all day, every day. Therefore, TVH opted for an intelligent solution - artificially intelligent, that is. They wanted to use artificial intelligence to automate the processing of these orders, thus putting the customer support's focus back on maintaining customer relationships and finding new customers.
The first step of this process was the analysis phase: going through their data, finding out what historical data was available and how it typically was processed. After the first phase, we started defining proofs-of-concept. Most importantly, we investigated the use of natural language processing (NLP) to make the process more efficient. This seemed to be the first step in the right direction, so we started to build this into a large, innovative application that is burned in TVH’s workflow.
The operational gain makes it possible for us to deliver more qualitative work with the same amount of people. Kristof Coudenys, CEO TVH Equipment
The final product? An application that offers full integration with TVH’s Office 365 mailbox, as well as their ERP software. The underlying AI models are continuously trained, based on new data that becomes available when the customer support corrects any errors via the intuitive user interface. Because of this, the application keeps improving and more valuable time becomes available. As a result, fewer errors are made and this reflects in the decreasing amount of credit notes.