It is a fact; manual document processing is time-consuming and slows down a company’s workflow.
Thanks to the progress of innovation, solutions for automating this task have been developed and are now increasingly used by companies. These solutions are based on the concept of Intelligent Document Processing (IDP).
In this new episode, Thilo Rossa, Chief Product Officer of Parashift, decided to interview Borisav Parmakovic, who has developed an end-to-end solution to end the headaches of processing specific types of documents: import/export customs forms and the likes with his company Digicust.
Let’s see how IDP can facilitate the processing of these documents 🙂
We have divided the exchange into two main parts. Feel free to click on each to get directly to the information you are interested in. 🙂
Part 1: Who is Borisav Parmakovic, and why did he create Digicust?
Borisav Parmakovic has worked for many large companies, such as DHL express. Often assigned to the customs department, he realized that employees faced many difficulties processing the documents. Indeed, when the processing is done manually, it is incredibly long and tedious because of the various and complex papers. So, he started thinking about a solution to relieve them of this task.
Some years later, he has the great idea to create Digicust.
Digicust is an end-to-end solution in the customs space that, thanks to the technologies used, optimizing the treatment process.
The whole concept is based on three robots:
1) The first robot is the basis of all: “Dexter IDP”
It is an end-to-end creation of customs declaration by applying Intelligent Document Processing (IDP) techniques combined with master data. It creates a wholly structured and electronic customs declaration.
2) “Taric” is the second virtual customs robot.
It does the tariff classification based on the data he gets from Dexter IDP.
3) “Neo” is the third one
It is a fraud detection robot, document control, and document request robot, which will be further exploited in the future.
“Taric” and “Neo” robots get all the structured data and can learn through the feedback of the customs specialists and indirectly from customs authorities and even customs officers.
“The goal of these customs robots is to create a cycle of qualitative and structured data so that the creation and completion of customs declaration are possible to end to end.” Borisav Parmakovic
All these robots are integrated into a pre-customs software called “Digital Customs”.
Part 2: Increase the IDP performance; a challenge for complex documents (like customs documents)
When people are looking for an IDP solution to get rid of manual processing, it is very often, to process invoices and other documents related to Account Payables. In this case, there are no more than two or three types of documents: the invoice itself, and the credit note for example. The image quality is very often good.
On the other hand, when the documents are more complex, the challenge increases by one level. Indeed, for the customs documents, there are very many types, also the quality is not very high because the paper documents are often swept a little everywhere before.
“In 20 to 30% of the cases, the documents are of poor quality, and 10% are not accepted (the users must then re-download them)”. Borisav Parmakovic
For example, customs clearance invoices are pretty complex, because the data points and all the fields from the document are complex.
As Borisav Parmakovic explains, if on an invoice we want to extract 100 fields, not all of those are structured. This requires a lot of manual validation efforts. Fortunately, Digicust has a customs machine that can create a customized declaration from all the data. So, it doesn’t take too long for the customs specialists to get the data. (At Digicust, teams need 5 to 120 minutes to create the customized declaration). The customer doesn’t have to do much. They upload the personalized file. Digicust and Dexter IDP do the rest.
This Intelligent Document Processing still requires a lot of human effort. However, with the help of their machines customs know-how, employees don’t need advanced knowledge of the subject matter, so they only focus on annotating the data.
Part 3: Find the sum-up of the whole discussion
To understand in more detail each theme discussed in this podcast, I invite you to watch the entire video (with the video is always better, but listening to a podcast while driving is not bad as well 😊)
Thanks to Borisav Parmakovic and Thilo Rossa for sharing their experiences and ideas on the subject.
Find the audio podcast now on Spotify, Apple Podcast, and Deezer.
Stay up to date with Digicust news
Stay up to date with Parashift news
Have a look on our latest episodes here!
