Bots with a Touch of Humanity

What Machine Learning Means to Automation Anywhere… Moving from Robotic to Intelligent Process Automation

For decades, companies have used business process outsourcing, offshoring resources, and business process management to automate and cut costs. The benefits of these solutions, however, are limited by their dependency on human labor, wage arbitrage, and capital expenditure needs. In this climate, corporations have started to invest heavily in Robotic Process Automation (RPA) – the application of software technology to automate traditionally manual, business workflows[1].

Automation Anywhere (AA) is a leading RPA solutions provider to Global 2000 corporations. Founded in 2003 and headquartered in San Jose, the company’s core product is an Enterprise RPA solution where “Task Bots” (software bots) are configured to automate repeatable, business process on most computer interfaces and software applications. Such processes include supply chain materials management, payroll processing, payment processing and database administration. As software is much more scalable, reliable and accurate than humans, AA’s customers benefit from immense cost savings and operational efficiencies. The consolidation of data from multiple tabs in Salesforce’s CRM platform, for example, could have taken an employee 15 minutes to perform. A Task Bot will only require 25 seconds and runs 24/7. With this magnitude of impact, it is perhaps unsurprising that the RPA industry is poised to reach $3.1bn by 2025.[2]

RPA technology, however, is still limited to automating clearly delineated business processes that digest structured data only. Yet structured data accounts for merely 20% of a company’s data. The remaining 80% is “Dark Data” – unstructured data that is difficult to assess, access and importantly, data that traditional RPA applications cannot process.[3]

As such, AA has invested significantly in machine learning to enable its technology to deliver greater customer value. Its “IQ Bot” – a cognitive software robot that digests semi-structured or unstructured data by learning from human behavior – marks an important step towards minimizing required human intervention in automation processes while expanding the scope of AA’s applicability. IQ Bots deploy machine learning in three steps. First, through computer vision, natural language processing and unsupervised learning techniques, IQ Bot can extract specific information from unstructured data in the form of images, emails, videos and scanned documents. Second, IQ Bot trains itself to make business decisions about this data by observing how humans have interpreted similar data in the past. Finally, by receiving human feedback on the quality of its outputs, IQ Bot undergoes immediate and constant self-improvement that refines its algorithms for identifying data more accurately and processing it to achieve the most optimal outcome. Customers can now increase the speed at which information can be processed and dramatically improve the accuracy of data digestion and the resulting commercial impact. The Intelligent Process Automation industry is expected to reach $8.3bn by 2023 and will fuel much of AA’s future growth.[4]

AA announced its IQ Bot 6 in March 2018. As it continues to develop and refine the solution’s machine learning capabilities, the company is becoming increasingly focused on maximizing the applicability of IQ Bots across more varied forms of data and different industry verticals. In the short term, AA is leveraging relationships with third party, pure machine learning partners such as IBM Watson, Google Cloud AI and Microsoft Cognitive Services, by providing these platforms a commercial outlet through its IQ Bot product. However, in the long term, AA is committed to building out its own, proprietary cognitive technologies to increase its defensibility against other emerging machine-learning driven bot solutions. In July 2018, the company announced a $250mm Series A investment, much of which will be used to develop “specialized machine learning capabilities and sophisticated AI integrations to drive higher operational efficiency, increased agility and flexibility to scale up-and-down anytime”.[5]

As the need for intelligent automation grows, AA should continue to invest in maximizing the speed of deployment, ease of integration and scalability. IQ Bots are trained and tested with data extracted from external software platforms. The solution needs to integrate seamlessly with a large ecosystem of software systems and applications to remain relevant and useful. Furthermore, its data extraction architecture needs to be low latency and high fidelity to scale effectively across an organization. Finally, AA’s technology is deployed by business users, not data scientists. Ensuring that the user interface and training process of IQ Bots is simple will become an even more important commercial point of differentiation.

Reflecting upon the impact of machine learning on business process automation and organizational structure, below are two questions to consider:

  • To what extent can a winning machine learning application remain generalized (i.e. applied across verticals and across a variety of use cases) in the face of competition from smaller, industry and application specific applications?
  • What are the ethical implications of redistributing / replacing human capital with an automated workforce and who bears the responsibility of such displacement?

(Words 791)

[1] Forrester, “The State of Robotic Process Automation,”, accessed November 2018


[2] Grand View Research, “Robotic Process Automation (RPA) Market Worth $3.11 Billion by 2025,”, accessed November 2018


[3] Courtney Kay, “Cognition and the future of marketing,”, accessed November 2018

[4] Cognilytica, “Intelligent Process Automation Market Report 2018,”, accessed November 2018


[5] Automation Anywhere, “Automation Anywhere Raises $250 Million, Reaching a $1.8 Billion Valuation in one of the Largest Series A Financing Rounds,”, accessed November 2018


Power to the People: How the City of Boston is driving civic innovation through crowdsourcing


Google Earth Engine: Organizing Geospatial Data to Save the Planet

Student comments on Bots with a Touch of Humanity

  1. So good to read about “the future of work” which should be/is top of mind for a lot of companies today. I see how, with the majority of a company’s data being “Dark Data”, it still quite difficult for RPA to be fully utilized and its benefits fully realized. IQ Bot sounds like a critical next step and it sounds like AA was essentially forced to make the investment in developing IQ Bot upfront – I imagine that companies were/are not willing to commit to AA software knowing it could only accurately process 20% of their data. But will IQ Bot be sophisticated enough soon enough for AA to get a proper return on this investment? Are IQ Bot’s data capabilities at 50% now? 60%? What’s the threshold required by companies today (who I’m hoping are sophisticated enough to understand that 50% data processing via software can still be worth it)?

    One of my biggest concerns would be around what happened with IBM Watson – the technology was phenomenal, but the requirements behind integration were too much for companies to bear, discouraging them from pursuing implementation of the software.
    Another concern I have is the one you mentioned about the ethics of displacing human workforce. Frankly, I’d be extremely disappointed if AA is not contributing to the dialogue around the ethics of human capital redistribution; while I can’t say I know WHO bears the ultimate responsibility for the implications of this workforce “transition”, I think the least AA could do (as a contributor to this workforce displacement) is foster collaborative and constructive conversation around the topic.

Leave a comment