The Regulatory Authority saves 80% of its employees’ time by using AI

How a government regulatory authority will save 80% of its employees’ time on a routine process by using AI

INDUSTRY: PUBLIC SECTOR

Sorting through emails and ensuring that they are routed to the correct recipient is an important and inevitable task for any organization. For this Danish regulatory authority, processing emails at the front desk has always been extremely time-consuming. However, by applying AI, emails can now be routed automatically, without compromising the accuracy and quality of the process. AI will transform this organization by routing 84% of emails through automatic processing.

How emailing can hinder an organization’s workflow

According to McKinsey, the average employee spends an estimated 28% of the workweek managing emails, which translates into more than 11 hours per week. What these statistics fail to include is the additional time it takes an employee to refocus on other tasks afterward. The lengthy process of reading, writing, and tagging emails can be significantly simplified through automatic processing so that instead, employees can focus on more valuable tasks – creating real value for an organization.

The challenge

Processing emails is a very time-consuming process. For this Danish regulatory authority, it is estimated that the front desk manually routes an average of 100 emails per day, which delays the organization’s workflow.

Developing an AI model with the ability to learn the rules of routing large portions of emails, can significantly reduce the time spent by employees handling this process manually. By applying an AI model to redirect emails to their proper receivers, the authority can decrease errors that may be made through manual routing, and employees can instead devote their time to more valuable tasks.

2021.AI is working together with the regulatory authority to:

$

Apply Natural Language Processing (NLP) to reduce time spent on emails by the front desk

$

Build a suitable and optimized AI model that can differentiate between inboxes with impeccable accuracy

$

Improve the organization’s workflow without compromising the quality of email handling

The solution

The aim of this project is to build an email routing model that can accurately predict the receiver of an incoming email. Email routing is a sophisticated and modern solution obtained by the means of NLP, which is when a document classification AI model learns how to classify emails into different groups. In this case, five main inboxes have been identified, each dealing with a different topic.

The model is trained on historical data to enhance its ability to predict an email’s destination.

An important aspect of this model is that the underlying mathematical algorithm applied does not understand human language, but instead, learns the structures or relationships of a text that is represented in a numerical format. Therefore, the first step is to remove linguistic noise and reduce text features, preserving only the useful information through the following methods:

: NLP preprocessing frequently used methods for cleaning linguistically messy texts

FIGURE: NLP preprocessing frequently used methods for cleaning linguistically messy texts

This preprocessed text corpus is then transformed into a term frequency-inverse document frequency (TF-IDF) numerical format, where each column is related to a specific word or sentence. The TF-IDF statistic is intended to reflect how important a word is to an email in a collection. A set of different mathematical algorithms are then trained and based on evaluations, the best performing model is selected for the task of routing emails at the front desk.

Once a suitable and optimized model that is accurately able to differentiate between the target groups is built, the model will be put into production and measure the positive effects of email routing. Some of those positive benefits will include increased productivity, a remarkable reduction in time spent on irrelevant emails, faster case response times, automatic tagging, and an increase in customer service quality.

Three key takeaways from this project:

$

The AI model can sort and route emails with 90% accuracy

$

The model can practically sort and route 84% of the incoming emails, leaving only 16% for the front desk to handle manually

$

By implementing an AI model, the authority can reduce as much as 80% of the time spent on emails

The results

By applying AI into the organization, employees at the front desk will be able to increase efficiency and productivity, as emails will instead be routed automatically without compromising the accuracy and quality of the handling. The lengthy process of manually routing 100 emails per day will be reduced significantly by automatic processing, which will route 84% of the emails.

The AI model can reduce the time spent by employees on emails, freeing up time to work on more important and value-creating tasks.

Get in touch

Interested in taking AI into production and scale to every corner of your organization?