In the Dynamic approach, Jiffy uses the concept of pseudo names and a built in repository. The documents are processed and values are extracted based on a dynamic repository which consists of a generic data model that defines the list of fields to be extracted as part of the process and a data dictionary which contains all the pseudo names for the fields listed in the data model.

In the Cognitive approach, Docube , the cognitive layer of Jiffy is introduced - The training data ( approximately 12 months ) is fed into the BOT and the ML model is generated . This model is then applied to the PDF’s being read

This approach uses an In-built Cognitive Engine based on Apache Spark and Multiple supported algorithms including Random Forest. One can achieve 85% straight through processing after learning if not more

Tags are created for all the fields from which data needs to be extracted for further processing. All the required data to be extracted as per the business use case has to be added as a Tag. For Eg: If reading an Invoice document - the typical tags created would be for “Invoice No”, Invoice Date”, “Invoice Amount” “ PO Number” etc.

Tags are the identifiers within the template through which JiffyRPA creates the model and Pseudonyms refers to the dictionary added to each tag from which the model identifies different formats of PDFs and extracts the relevant data. For eg. for a tag of Invoice number – the various pseudonyms one could possible create would be
“Invoice No”
“Inv #”
“Invoice Number”
“Reference Number” etc.

Rule Based Approach is one of the ways to process the documents. In this approach JiffyRPA identifies the fields that exactly match with the provided template. Every template in the application is identified by the tag provided. The PDF node identifies the template through the tag provided

Rule Based Approach is one of the ways to process the documents. In this approach JiffyRPA identifies the fields that exactly match with the provided template. Every template in the application is identified by the tag provided. The PDF node identifies the template through the tag provided

This activity should be ideally done jointly by the BOT designers and the Business/functional team to achieve best results. The business team has the best functional know-how and would be able to work efficiently with the designer to define the data repository

We create a model for each field and for each field we use binary classifiers. We can also use multi classifiers. But binary classifiers have given us better results.

The Jiffy Data Interface (JDI) component of Jiffy helps in identifying and addressing exception scenarios in the automation process A user can view the status of each transaction that is being performed by the BOT on JDI based on status that is assigned to the record at every phase of the process. Any exceptions encountered during the process are highlighted on JDI which can be customised based on action items that the team wishes to perform on these exceptions. The user can decide to correct the information which resulted in the error and resubmit to the BOT for reprocessing. The exception can be managed by sending an email to the concerned team and informing the concerned users of the details of the exception.

The design of Jiffy is also in a way that every single node has a status that is returned on the execution of that node which in turn can be tracked and customised to perform the next action based on a success or failure. Additionally, analytical dashboards can be easily created in Docube with interactive Visuals