Facebook IconFacebook IconFacebook Icon

Automated Email Processing

Sagacify Logo

Challenge

Companies selling services directly to consumers such as insurance and banks are often faced with an important amount of inbound emails. To process these in time, some collaborators are dedicated solely to distributing emails to the right person according to the customer's requests. These employees look at the subject of the email, the content of the email and the attachments, in order to determine who should treat the email, then route the email to the right person or department, or encode it into an document management system, linking it to the right client or file.

DAS, and other insurance companies, came to Sagacify with the objective of freeing their collaborators from this repetitive task in order to let them focus on providing the best service, especially in complex situations. 

The right AI system, integrated in companies’ inbound customer email management can increase operational efficiency, quality and revenue.

By focusing their team on the most priority and important cases, companies can treat their client's emails faster, more accurately, and at a lesser cost.

Solution

To automate the email processing tasks, we trained, developed, and deployed models that automates the analysis of emails and assigns several labels to it, such as the type of communication, the level or urgency, the identification of  the client, the type of attachment, etc.
Our solution finally uses the extracted labels to forward the email to the right person or department, link the email to the client or file in the database, and can even propose a set of most likely answers to the operators.

Key elements  

  • Custom model trained to maximize classification of the emails according to the taxonomy of the client.
  • Custom model trained to extract specific information such as client number, contract number, … 
  • Scalable and secure system deployed in production, which manages the entirety of incoming emails to the insurance, over multiple mailboxes. 
  • Live monitoring of the performances of the solution and its performances
  • The model continuously collects feedback on its performances and is regularly retrained to improve.

Results

Today, almost 100% of the incoming emails treated by our email solution are automatically processed without need for human intervention.

Employees that were previously in charge of this classification can now spend more time on valuable and fulfilling tasks.

Related cases