Our Client, a leading financial institution, realized that the manual process of classifying internal issues e-mail and forwarding them to the corresponding units was slow, inefficient and cumbersome. All e-mails were sent to a central “Help Desk” that manually classified the need and put it in a formal ticketing system, forwarding the e-mail to the corresponding department. The Client was looking for ways to improve this process.
With this goal in mind, Open Web Technology was asked to provide an Artificial Intelligence system able to pre-classify these e-mails. The idea was not limited just to the classification but also to provide enhanced features for prioritization and context information.
Open Web Technology's team, together with their client, defined a four-step approach to turn this vision into reality:
Our team, in close cooperation with the project sponsor, was able to identify the scope of the project and define the desired functionalities and requirements for different use case scenarios.
In order to process thousands of e-mails, our team developed a robust agent capable of receiving and extracting information from a mailbox, interacting with the AI agent and redirecting the enhanced e-mails to the corresponding outbox.
In a first step, an initial set of both structured and unstructured e-mails was provided to train the AI system under supervised learning. Advanced AI techniques such as Named Entity Recognition, Boosting, Keyword Extraction, Pattern Recognition were used to do the classification and enrichment.
In a second stage, the training switched to real-time processing and classification of the e-mails for further improvements. Context information was extracted from the e-mails to provide insights on the e-mail and its sender.
Feedback and Tuning
Through close cooperation with the project sponsor, we automatically received feedback into our system to efficiently tune the AI agent, reduce the error rate and tweak the system.
Thanks to the work of Open Web Technology and the possibilities of Artificial Intelligence, the Client was able to drastically decrease the processing time of tickets, reduce operational costs and downtimes, while increasing productivity. The addition of context information to the classified e-mails provided great value to the Client