Modern Problems Require 'Intelligent'​ Solutions

Modern Problems Require 'Intelligent'​ Solutions

Dave Chappelle's famous skit aired on January 21st, 2004. As if in response, a company called DeskOver is formed a year later and begins by providing automation solutions to various Banking, Financial Services and Insurance (BFSI) companies. These solutions were software libraries that could be embedded in the company's software. A little ways down the road in 2012 after they had been working with IMB, they realised that these modern solutions could be developed into an end-to-end Robotic Process Automation (RPA) software. They re-established under the name UiPath and became the industry leader in RPA software.

Zoom in to present day. RPA has soared in popularity, However, as most Business Analysts (BAs) will know, two key criteria that hold a large influence on the feasibility of implementing an RPA project are:

  1. The amount of Unstructured Data; and
  2. The amount of time until an expected system change.

Why? It's easy to implement an RPA strategy when it's known that every piece of data needed is found in the same 'place', in easily scrape-able form. For example, a company needed to get home addresses from all of its employees contracts, however these contracts are structurally different depending on the year you signed the contract, on your job role, on your seniority within your department, etc. Also, some of the documents are scanned in as images, some are PDF documents, some are word documents.

This variation in processing doesn't lend itself to RPA in the traditional sense. RPA is useful for automating those repetitive and time consuming tasks, and the solutions are made to work with GUIs of web and desktop applications. However, when an application has updates, then the RPA design may need to be redesigned, as the GUI will be slightly different, or the CSS or HTML has been altered and this may have an effect on the selectors used in the design.

Welcome Intelligent Process Automation (IPA), the first steps in adaptable, artificially intelligent RPA. Automation Hero, Inc. unveiled their AI Studio software to the public this week. Their 'no-code' platform allows users to design AI models that can be incorporated in their Hero_Flow RPA design studio. Most notable features include:

  1. Intent Detection: discovering the intent of incoming emails, a source of unstructured data.
  2. Reinforcement Learning: a type of machine learning where the output of the current iteration of the process is used as an input for the next iteration. The 'agent' repeats this 'trial and error' process, and finally will output the best solution.
  3. Dark Data Extraction: a tool to extract structured data from unstructured data. This area of data analysis is seen to many as the basis of the modern industrial revolution.IBM estimates that 90% of data generated by sensors and ADCs is never used, leaving only 10% of clean and structured data. It's easy to see that in a time of 'Big Data' companies are keen to maximise extraction and reduce their wasteful processes. After all, data centres require resources to be powered.

It's exciting, it has massive potential and it's definitely going to help us at Accelerate RPA to continue to find 'Intelligent Solutions' to our clients 'Modern Problems'.