When IA meets AI!

July 29, 2019

Circa early 2000s: An Accounts Payable staff would come into the office every day, read his or her emails, open the attachments (MS Excel, PDFs or others) and manually copy relevant data from these attachments into an ERP system.

Circa 2017: An Accounts Payable staff would come into the office every day, read his or her emails, opens the attachments (MS Excel, PDFs or others) and manually collates the data into a single format. Then the bots take over, validate and insert the data into an ERP system.

Cut to today’s date: An AI-driven bot first collects and consolidates all the information from varied sources (different kinds of Excel files, PDFs, scanned images, etc.) and then also copies the relevant information into an ERP system. The Accounts Payable staff is reviewing the process regularly and making necessary corrections, if required.

Most of the data, approximately around 70-90%, is dark data (semi-structured and unstructured). Most companies are still dealing with PDFs – these could be contracts, invoices, purchase orders and so on. If you are a large manufacturer, chances are you will have thousands of suppliers, each with a unique invoice layout! This semi-structured data is hard to process by rule-based systems, but if you bring AI into play you can deal with this semi-structured data.

Today is the day and age of combinational Artificial Intelligence and Intelligent Automation (fully supported by humungous computational power). The maturity of AI / ML and its ability to tackle a wide range of business problems, courtesy the advances made in Natural Language Processing, Object Recognition, Document Analysis and Classification, have made end-to-end automation possible. Just in the Document Analysis space, what OCR could do with at best 50-60% accuracy, an advanced AI-driven system could get to 90-95% precision, by using supervised and non-supervised machine learning.

The limitation that RPA had, all this while, was around decision making. The rule of thumb has been that for any rote or repetitive tasks, RPA is the way to go. But now, with the power of AI / ML, even decision-making and intelligence can be automated. This is more so in the case of unified tools that the likes of Automation Anywhere provide.

This is where things become enterprise ready – RPA task bots for rote tasks, IQ Bot for intelligent data extraction, and Control Room for monitoring / managing all the bots.

We can now truly say that RPA is ready to solve much bigger industry-grade problems, ushering us into the fourth industrial revolution – where software automation will be strategic, transformative and easily scalable!

Not convinced yet? Then the following forecasts should do the job – according to a Forrester report, the market for RPA will be valued at $2.9 billion by 2021 – a massive jump from the $250 million reported in 2016 (

At RoundSqr, we believe that the next big thing is convergence of Intelligent Automation and Artificial Intelligence. This will open up new frontiers, creating a state-of-the-art Digital workforce!

Let us take out the robot from the human:)