EDDIE Case Study

InAssist and Deep Data Insight work together to produce fantastic results using the EDDIE Artificial Intelligence platform



Deep Data Insight has created an Artificial Intelligence factory that produces bespoke and licensable AI solutions for organizations from almost any sector.

One of their modular platforms, EDDIE, uses Optical Character Recognition and Intelligent Character Recognition technologies to provide huge ROI for companies that have medium to large amounts of data to process.

InAssist is an insurance broker based in California. They deal with a multitude of insurers, making sure that their clients receive the best policies available to suit their needs.

You can read more about OCR and ICR here (insert link to OCR/ ICR blog)



Like many industries, the insurance sector relies heavily on individual pieces of data. These are usually in the form of claims, and the data can arrive in many shapes and sizes.

The job of an insurance broker is to represent a number of insurers…so it follows that whilst a single insurer may have thousands of individual claims, a broker will be dealing with an exponentially increased amount of information.

A broker can never rely on how this information is received. It will range from typed and printed documents to hand-written notes to PDFs. They will often include images and tables as well as text.

Nevertheless, all of this information needs to be processed with extremely high levels of accuracy – a simple mis-step may result in a customer’s claim being nullified.

The challenge detailed above was certainly one that InAssist recognised. There are eleven different sites, with every site sending in their claims in different formats. Each claim needed to become part of a trail of documentation that would always ‘belong’ to the same insurer. And, they all needed processing!

Like most brokers, this had always been a manual task for InAssist. This meant dozens of administrators whose role it was to ensure the accurate and expeditious processing of the claims onto a master database. Often, each claim would need to be typed out because it was unclear. And with each keystroke, there was the potential for errors.



OCR forms part of a technical solution for exactly this problem. OCR is a technology that enables scanned documents to be digitised very quickly. OCR will identify typed letters and form them into words, so that a document file can be opened for each case. So, if a claim has arrived in PDF format, this document and all subsequent data from this insurer and customer will be processed into the same digital file with great accuracy and speed.

Furthermore, ICR will solve the same problem for information that has arrived in cursive, or hand-written style. ICR has the ability to decode a hand-written note, then store the handwriting so that all other communications in the same handwriting can also be identified and correctly filed.

EDDIE has the ability to apply a percentage ‘risk rating’ to each individual character it processes. So, if it is uncertain about something because it appears to be blurry, it will flag this to a supervisor. And of course, it will learn this rule for the next time.

EDDIE will also correct any mistakes as they appear. So, if the same individual has sent in two forms, one of which has a mis-spelled address, EDDIE will flag this and if needed, correct it.

This is the beauty of EDDIE. A machine-learning tool that will accurately extract information into a common format and quickly sort individual claims, replacing the human administrator who will take time to do the same job, and will not do it as accurately.

InAssist worked with a number of insurers, all with their own formats and models. EDDIE recognises which format is from which company, and files accordingly.

Of course, once this information is in the digital domain, it is extremely easy to find it and compare it with new information as it too arrives.



EDDIE has been running for over three years now on this project. It required relatively little set-up, and once in situ hardly any maintenance. 

Since EDDIE is customisable, it has been programmed to fit in perfectly with the InAssist system.

The company manages its workflows around EDDIE; EDDIE will process the documents overnight, so that they are ready for when staff arrive in the morning. Staff then merely need to cast a quick eye over the reports from EDDIE, and continue with their work.

Within InAssist, EDDIE is easily processing 11,000 files per month, which is roughly equivalent to the work rate of a team of ten. And, with better accuracy. This provides an incredible ROI for InAssist.


What next for EDDIE?

This is just one of many applications for the OCR/ ICR tech that EDDIE uses. Other applications include the deciphering of Doctors’ handwriting for prescriptions, and use within the prison system. Here, different drugs and packaging for different convicts were loaded onto a conveyor belt for checking prior to distribution. EDDIE would detect the wrong colour drug – for example paracetamol should always be white – or even the wrong shape. It would read the labelling too, and report if there were any irregularities for an individual’s treatment.

Jeewa Perera is the CEO of Deep Data Insight, and has commented: 

“EDDIE can provide massive return on investment wherever there is a process that relies on medium to large amount of data.

“Not only can EDDIE do the job of a number of humans, but it can carry the task out with high levels of accuracy, making workflows more and more efficient.

“EDDIE has a fantastic user interface and reporting tool. We are proud to have created a system that needs little or no training to become proficient in. This means that the job of data processing can be handled with little fuss and at relatively small cost”.

For further information about EDDIE, follow this link or complete the enquiry form on this page:                                                https://www.deepdatainsight.com/case-studies/rddt-case-study/