automating the reading of paper forms

The adoption of Intelligent Character Recognition to automate document management

Despite the diffusion of the digital world, part of the work continues to be done on paper in the insurance, banking and financial sector. In particular, this concerns specific customer segments or procedures for which it is still necessary to rely on old methods where paper forms are filled out. For this reason, a huge volume of paper remains. This requires large storage spaces together with high resources and costs, both in terms of data processing and maintenance activities. This activity is therefore often entrusted to outsourcers, with the aim of reducing the burden on internal processes, or at least containing their costs. How can we intervene in this process and overcome its current limits using Intelligent Process Automation technologies?

THE DOCUMENT CONTROL PROCESS

A document control process typically consists of five specific steps:

  • strip-0

    DOCUMENT COLLECTION FROM SEVERAL CHANNELS:

    during this phase, the documents obtained through various channels such as fax, e-mail, uploads or document systems are collected by a single orchestrator system, interpreted by a robotic component.

  • DOCUMENT CLASSIFICATION:

    on the basis of their attributes, structure and content – consisting of text or images – each document is classified to make it possible to manage correctly the data extraction workflow and carry out any validation that might be required by human operators.

    strip-1
  • strip-2

    EXTRACTION OF RELEVANT DATA:

    the relevant information is extracted based on the type of document; this may include dates, amounts, identification codes, IBANs, ISINs, policy numbers and any other data that may be needed by the robot to carry out its actions.

  • CONSISTENCY AND CORRECTNESS CHECKS:

    the extracted data is checked to assess the consistency, completeness and correctness of the individual pieces of data and, where appropriate, in relation to other documents.

    strip-3
  • strip-4

    EXECUTION OF OPERATIONS ON VARIOUS SYSTEMS:

    once the data has been extracted and validated, it can be used as part of the automatic robotic processes, thus relieving the human workforce from the burden of data entry on the various systems involved.


THE BENEFITS OF INTELLIGENT CHARACTER RECOGNITION

Sprint Reply has focussed its solution on specific types of activities characterised by high repetitiveness and standardisation, which can therefore benefit most from the introduction of Robotic Process Automation tools.



First, the forms were updated, redesigning them and their general structure. Thanks to this approach, the document templates were designed according to the technological tools, while at the same time the manual data entry was made more practical and better controlled. In defining the new forms, specific data extraction rules were also taken into consideration, thereby guaranteeing the best possible use of the Intelligent Character Recognition technology.

This has made it possible to obtain a fast and efficient processing of the information contained in the forms, which can easily be transformed from analogue data into digital data. Once confirmed, the data is automatically entered by a robot into the bank’s or the insurance company’s systems, without the need for further interventions. Any anomalies or discrepancies can be easily intercepted, thanks to the verification procedures, and are then directed to an operator who will manage them accordingly.

Thanks to the use of advanced technologies, it therefore becomes possible to rethink the need to outsource the procedure for checking and processing paper documents, in favour of an almost fully automated and digital workforce. In the transformation brought about by the new automated procedure, the savings in terms of time and the economic advantage prove to be significant. Activities that previously took internal staff more than 10-20 minutes to complete are now managed fully automatically. Moreover, specific procedures such as, for example, the verification of Privacy documents, previously outsourced at a cost of more than 2-3 Euros per file, are now fully robotised, with success rates above 90%.