From OCR to Intelligent Document Processing: a new era for processing documents automatically
Given today’s challenges related to digitalising customer relations, coupled with necessary productivity gains, automated document processing now plays a strategic role. New solutions are emerging, from OCR to the latest advances in machine learning and artificial intelligence. Frédéric Le Bars, Innovation Director at ITESOFT, talks about this revolution already in progress.
OCR, IDP, NLP... a few terms you need to know:
- OCR: Optical character recognition
- IDP: Intelligent Document Processing
- NLP: Natural Language Processing
Document digitalisation has started a revolution.
Frédéric Le Bars: The paradigm has changed. With today’s new uses for digital technologies, increasing numbers of customers want to send their documents to their bank, insurance, mobile phone carrier, and others, easily and in real-time.
Organisations therefore need to revisit their relational model and offer interaction via new channels that are ever-faster, easier, and more digital to engage customers. But it is not that simple.
What does it involve, in practical terms?
FLB: According to a survey by ITESOFT-L’Assurance en Mouvement published in 2022, 75% of the organisations surveyed now offer portals or applications that enable customers to submit their documents online in sectors such as insurance. However, in more than half the cases, these interfaces do not integrate intelligence to analyse those documents in real-time and confirm that they are properly taken into account in the customer’s file.
The result can be a frustrating customer experience with people giving up in the middle. The organisations reaching out to ITESOFT have understood that without intelligent document processing, you cannot have a successful customer experience.
How do new document processing solutions resolve these issues?
FLB: When a customer submits a file or document on a portal or web application, what they really want is to receive real-time feedback: Did I send the right document? Is the document valid? Is my file complete?
That is what we do with our robots, which analyse all new documents on-the-fly. They identify the nature of each document, check compliance, extract relevant data, and also verify file completeness. Customers are guided in real-time along a seamless journey. At the same time, productivity is improved for back-office teams.
“Without intelligent document processing, you can’t have a successful customer experience.”
Frédéric Le Bars, Innovation Director at ITESOFT
Concretely, what new technological challenges does this represent?
FLB: From a technical perspective, the challenge revolves around how to process more complex documents faster. Paradoxically, the documents received today are actually less good than before, whereas the stakes of processing them rapidly have never been stronger. Documents that used to be scanned at 300 dpi are now being replaced by photos taken by customers using their smartphones – documents that are often less well lit, sometimes folded, wrinkled or even stained...
To handle this new complexity while still providing immediate response times, document processing technologies must become smarter, no longer just based on OCR, but increasingly with real-time analysis of images captured by cameras. This new era of document processing is called Intelligent Document Processing (IDP), which leverages artificial intelligence and particularly Deep Learning algorithms.
How do you see Intelligent Document Processing?
FLB: Our vision of IDP involves mastering all the relevant technologies - OCR, expert systems, Deep Learning, Natural Language Processing, and other AI algorithms – and knowing how to combine them intelligently to take full advantage of automation in all use cases, such as paper, scans, photos, structured and unstructured documents, and more.
For standardised documents such as supplier invoices, OCR and expert systems tend to be the most effective. On the other hand, when variations in the nature and quality of documents increase, which is the case when sending a smartphone photo of a letter, ID card, or car registration, it is essential to use our specialised AI for image processing, information extraction, and natural language understanding to offer an optimal interactive customer experience.
Our ability is knowing how to combine each of these technologies in a unique manner that makes our solution different and guarantees that our customers enjoy the highest possible level of performance, regardless of the type of documents and their quality.
“Real intelligence is being able to master all the relevant technologies – OCR, expert systems, Deep Learning, Natural Language Processing, and other AI algorithms – and know how to apply them at the right time according to the context." - Frédéric Le Bars, Innovation Director at ITESOFT
Does using more AI make it more complicated for companies?
FLB: Quite the opposite. Our vision is to handle all the complexity on our side. The companies using our solutions do not have to configure or optimise the various AI components. Our experts take care of everything.
One of the reasons we developed the largest document knowledge base was to make it easier for companies to access these technologies. For marital status, revenue, contact information, identity, bank details, health information and much more, our solutions integrate over 130 pre-configured documents that are optimised constantly and above all, available right from service launch.
Basically, the promise of this new intelligent document processing era is to handle all that at the same time with better performance and greater simplicity. Thanks to these innovations, organisations can finally fully transform their customer experience.
Communication and editorial director
Laurent joined ITESOFT in 2000. Marketing Project Director and graduate of IAE Aix en Provence, he has more than 15 years of experience in the production of events, studies and content around dematerialization and process automation.