Stop wasting valuable time on shared mailboxes

Many of us feel strained by work emails. Particularly group mailboxes engender a never ending stream of mails coming from customers or suppliers with simple knowledge inquiries or tasks. Typically, a set of employees is burdened with the task of handling these emails. Often this implies that employees have to sort through the inbox reading emails and categorizing them using a folder structure. Other colleagues then process emails from these folders by executing repetitive and burdensome tasks. Simple solution? Intelligent Mail Automation (IMA) powered by Ciphix. Implementing IMA allows for an automated process to perform all above-mentioned tasks at lightning speed.

What Intelligent Mail Automation can do for you?

  • Automatically categorize incoming emails
  • Understand requests & extract relevant data from unstructured emails
  • Auto fulfilment and response – instant reply, forward, escalate to process owner, call API & more..
When do you need Intelligent Mail Automation?

  • Your shared mailbox receives over 50 emails per day
  • It takes a lot of time to handle these emails
  • You have no overview of all the emails we are receiving


Accounting Services reaps benefits of IMA

IMA is an advanced solution that, for example, is implemented for one of our clients active in the airline industry. For their Accounting Services department, the solution automatically reads and classifies emails in their inbox. “New invoice”, “Reminder” and “Dunning letter” are examples of such classifications. Besides classifying, a fundamental feature is the extraction of entities. These two pillar functionalities constitute a solution that allows a process with unstructured textual data and complex decision making to be fully automated.

How does it work?

First up is reading and understanding the content of an incoming email. This process is called classification, a type of machine learning problem where the category of the email is determined by selecting one from a predefined list of options. For example, an incoming email may be assigned the category “Reminder”.

Once an email is categorised, the next step is detection and extraction of important parameters known as entity extraction. For example, for emails containing invoicing reminders, the extracted entities could be invoice number, issue date, and invoice amounts.

The information that is extracted serves as an input for automation of the remainder of the process. Thát is where RPA comes in.

RPA x AI – A match made in heaven

The extraction of entities is what transforms a difficult to handle input format into structured data that allows us to reap the benefits of Robotic Process Automation.

In the IMA solution, the extracted entities are selected based on the repetitive proces that comes next. For example, for the Accounting Services department, the next step was to retrieve a status update on invoice reminders. So that’s what we did.

Using the invoice number extracted from free-form text, we built an RPA solution. This RPA bot searches through back office software and gathers relevant information about the outstanding invoice. The remainder of the entities contribute to the reliability of the solution by allowing for additional checks to make sure the right invoice is matched. The output of this process can then be used to update the employee, or better yet, perform the follow-up action automatically. For example, we can use this information to automatically reply to the supplier by sending them an update on the status of their invoice.


Close cooperation with the dedicated Accounting Services team allowed us to build an AI that reaches mind-boggling results: more than 95% cases are processed correctly, without any form of human intervention. With an implementation time of mere weeks, about 160.000 emails can be processed per year, more than 300.000 entities are extracted and several FTEs of tedious work is eliminated using the IMA solution.

All in all, this project lead to a solution that is effective, efficient and reliable. Humans can refocus on work that matters and stop wasting time on a mailbox.

What is required?

One word: data. To build the machine learning software specific to the case, we start off with a batch of labeled historic data to teach the algorithm the complex decision making it needs to be able to do.

Feeling inspired to rid your team of shared mailbox duties too? Send us an e-mail.