Cognitive Robotic Process Automation adds artificial intelligence capabilities to traditional RPA tools. This enables RPA systems to learn and adapt to new situations and improve business processes. Businesses that adopt cognitive RPA can realize cost efficiencies at scale, become more responsive to customers, and ultimately redeploy employees to higher-value activities.
Despite the benefits offered by intelligent automation, a recent survey by IDG revealed that under 50% of US and European enterprises have deployed intelligent automation. That signals both untapped potential and underlying challenges to adoption, which we will explore throughout this post.
We will cover the difference between regular and cognitive RPA. We’ll then look at two leading cognitive RPA vendors, UiPath and Automation Anywhere. Finally, we will consider key takeaways for business leaders tasked with implementing cognitive RPA solutions.
Robotic Process Automation is not AI
Robotic Process Automation (RPA) software automates manual, rule-based and repetitive tasks that are usually handled by employees.
RPA is often used for document processing activities and can help companies save time, free employees from simple repetitive tasks, and focus on more strategic initiatives.
Most traditional RPA tools on the market do not use artificial intelligence. RPA software simply replicates the steps that an employee performs to generate an invoice or a report, for example. The employee’s actions and mouse clicks are recorded and replicated.
Crucially, this type of RPA only works as long as the process never changes. In most cases, the RPA tool must be updated when an invoice layout or reporting requirements change, for instance.
Cognitive RPA Incorporates AI
Cognitive RPA can integrate machine learning, natural language processing (NLP), computer vision, and optical character recognition (OCR) capabilities to RPA tools. Instead of simply replicating human actions, cognitive RPA can find the most efficient way to automate a task, process data from multiple sources, and deal with new data and changing requirements.
Machine learning enables cognitive RPA tools to adapt to new situations. For example, if a payroll system’s user interface changes, the cognitive RPA tool can recognize the changes and achieve the same result. It can also streamline accounting processes by identifying unaligned processes, duplicated activity, and irregularities. It can then help build an automated workflow that saves time and money.
NLP adds text processing capabilities, enabling cognitive RPA software to read, understand, and generate text. For example, it can recognize that the client’s name in a new invoice appears in a different section compared to previous invoices. NLP can also be used to extract relevant information from large databases to generate reports, invoices, or other documents.
In an insurance setting, cognitive RPA using machine learning and NLP can help automate the insurance application review process. Large numbers of historical insurance applications can be fed into a machine learning model, with each carrying a label of ‘accepted’ or ‘rejected’. The model then ‘learns’ how those decisions were made by ‘reading’ through the applications. The cognitive RPA model can then automatically accept or reject future insurance applications.
This enables insurers to respond faster to customers while saving time and money. More importantly, it allows insurers to reallocate resources to revenue-generating or product innovation activities, which clearly benefits long term growth, customer acquisition, and market share.
Deployed properly, cognitive RPA not only reduces costs and saves time, but also allows companies to dedicate more resources to growth and innovation.
Company Profile: UiPath
UiPath is a New York-based RPA solutions provider. It is also a unicorn (valuation over $1B) in the AI startup space.
UiPath offers cognitive RPA software to companies across industries, including banking & finance, healthcare, and retail, for example. Their products automate processes related to accounts payable, insurance claims, and legal work, among others.
Their flagship UiPath Enterprise RPA platform offers both regular and cognitive RPA tools for companies to automate processes at scale. Described as an end-to-end automation platform, it helps companies:
- Plan RPA implementation by identifying the most productive automation opportunities
- Deploy RPA tools worldwide and manage them from a central location.
- Run RPA tools that integrate into the existing tech stack
- Measure improvements and track KPIs to quantify the benefits of RPA
Employees across the organization can use the platform to collaboratively develop and deploy RPA tools. The UiPath platform also integrates with other popular enterprise systems such as SAP. It also offers AI capabilities such as machine learning, computer vision and NLP, as mentioned in demo videos on their website.
Computer vision is used to recognizes objects and images in the on-screen user interface, allowing RPA bots to adapt to user interface changes. Neural networks are most likely used for object recognition.
NLP and optical character recognition can process and extract data from high volumes of receipts and invoices that come in different formats. This would help in streamlining accounts payable and expense management processes.
UiPath also lets users make corrections to the bot’s work. For instance, if a bot extracts the incorrect vendor name from a PDF invoice (skip to the 2:36 mark in this video), an employee can point out the correct name and the machine learning model will use this input to improve itself.
These functions offer clear benefits for UiPath clients. Clients who invest the time to design tailored automation workflows and deploy the platform across departments will realize time savings, productivity gains, and even gain actionable insights from their data.
Company Profile: Automation Anywhere
Automation Anywhere is a California-based cognitive RPA company and is also a unicorn in the AI startup space.
The company provides cognitive RPA software and consulting services to multiple industries including financial services, healthcare, telecom, and manufacturing, among others. Their solutions are primarily related to document processing and workflow automation. Their clients include the likes of Sprint and Hitachi.
Automation Anywhere’s flagship product, Enterprise A2019 (latest release), integrates RPA, AI and machine learning into one web-based platform. According to an Automation Anywhere blog post, Enterprise A2019 allows a company’s RPA developers to work closely with data scientists — two functions that often operate in silos.
Enterprise A2019’s graphical user interface makes it easy for non-technical staff to create and deploy AI-enabled RPA tools. Automation Anywhere has also launched a bot app store where client companies can purchase crowdsourced AI bots for specific purposes (e.g. recognize text from videos and images). This makes it easier to deploy intelligent automation tools and overcome one of the main challenges of cognitive RPA adoption — that the tools are hard to build.
A key component of Enterprise A2019 is IQ Bot, an AI-driven tool that automatically reads and processes data from complex documents and even emails. IQ Bot allows users to easily call upon solutions for speech recognition, language processing and computer vision from third-party vendors such as IBM, Google, and Microsoft. This allows businesses to quickly process and extract insights from unstructured and semi-structured data, and achieve end-to-end process automation.
According to a case study, a global bank used IQ Bot’s machine learning capabilities to automate numerous HR functions while saving $1M annually. The bank’s Human Resources department had to manage hundreds of HR onboarding forms in multiple languages across many countries. The HR team used to manually sort, manage, and process these forms and manually enter data into the HR Management System. This approach was slow and prone to high error rates.
Automation Anywhere’s IQ Bot was able to handle high volumes of semi-structured data in these forms. First, task bots downloaded the forms and fed them into IQ Bot. Next, IQ Bot extracted and cleaned relevant data points from the forms and automatically entered records into the bank’s HR Management System.
After adopting Automation Anywhere’s cognitive RPA solutions, 40% of the bank’s HR Management System’s volume was fully automated with a zero error rate.
Takeaways for Business Leaders
Implementing cognitive RPA solutions in an enterprise can take time, requires collaboration between departments, and needs a clear firm-wide strategy. Executives tasked with driving these initiatives should ask themselves three questions:
- Where will we gain the most value from automation?
- Should we build in-house or work with third-party vendors?
- How can we implement it at scale?
First, to see where automation will add the most value, executives must understand their most pressing business needs. Are they in a mature and competitive industry where cost control is essential? If yes, they may choose to focus on automation initiatives that purely cut costs (e.g. compliance processes). Alternatively, if the industry is growing and competitors are rushing to gain market share, executives may choose to automate processes related to customer onboarding, for example. After all, the easier you make it for customers to join you, the faster you grow.
Second, the decision to build or buy cognitive RPA solutions (and AI solutions in general) largely depends on the firm’s internal AI and data science capabilities. Relatively few firms will have large numbers of specialized staff and the AI maturity to build everything in-house. At the other end of the spectrum, companies with little expertise can purchase cognitive RPA platforms from vendors such as UiPath and Automation Anywhere and implement individual RPA solutions before scaling up.
A hybrid approach might bear more fruit in the long term. Having internal subject matter experts and developers collaborate with cognitive RPA vendors can result in tailored, scalable solutions that are meeting the precise needs of the organization.
Finally, the goal should always be implementation at scale. That is where the highest return on investment comes from. It doesn’t make much sense for different departments to use independent RPA tools. The cost savings and data insights will be far greater if an RPA system can access data and optimize interconnected processes across departments and countries. Insights from systems that capture company-wide knowledge will help executives make informed, timely, and data-driven strategic decisions.