There has been plenty of confusion in the marketplace when looking at automation software. Even the names used to describe things seem to multiply like rabbits. These terms are developed and proliferated by analysts, software vendors, and solution integrators as they each try to provide their own spin on the market. So, let’s take a quick look at some of the terms that are used so we can create our own “decoder-ring” to make sense of it all.
Robotic Process Automation
This term was an early starter in the automation space. Robotic process automation (RPA) used robots, or “bots,” defined as software agents that would interact with applications just like a person would. Without defining specific programming interfaces, a process-analyst could define which parts of an application a process used and then “train” the robot to submit the changes according to a set of rules.
The applications could be based on Windows applications, web pages, mainframe applications, Java apps, or even home-growth applications that were written using long-extinct technology platforms. These bots would follow rules to interact with any of these applications, either simple rules such as “create a report and email it” or complex rules that involved many steps. These steps might include evaluating specific fields in the application and then following those rules accordingly, such as “Check the balance of the inventory. If it is less than a certain amount, issue a warning e-mail, otherwise process the transaction and initiate the transfer to the location listed on the purchase order.”
RPA is the foundational layer upon which intelligent automation and hyperautomation are built. These concepts require an RPA platform to permit interaction with the applications without programming the interactions. Without RPA, automating communications would require many new connectors to enable artificial intelligence (AI) to pull data from and initiate actions to the various corporate systems that contain the information required for processing.
Sometimes referred to as cognitive automation, intelligent automation (IA) links artificial intelligence with the interactive capabilities of RPA. The two basic concepts that intelligent automation links together are thinking and doing.
RPA is great at the doing and has several capabilities to manage work through rules, but there are some aspects of executing work that require thinking before an action can happen. Some of this thinking work involves reading documents, using optical character recognition (OCR) to pull out data into a form that the computer can use. Then intelligent document processing (IDP) can be used to understand the document type so it can be processed appropriately.
For example, anyone who has been engaged in processing invoices will tell you that nearly every company has their own format for sending an invoice to their customers. Because of this non-standardized way of sending invoices, companies typically employ people to read the images of these documents and type the information into the accounting systems.
Depending on the number of invoices, this might require a large staff or a contract with an outsourcing firm to keep up with the volume. By building intelligent automation to read and understand the documents using IDP then processing the results appropriately with RPA, the solution can reduce the amount of time that people are engaged in manual efforts such as typing information into the system.
But intelligent automation can be used for so much more than invoice processing. For example, by engaging a natural language processing (NLP) platform, an automated process can read emails that provide insight into questions being asked to a support team or to a chatbot that is allowing customers to interact in real-time with the company.
The NLP platform can understand both what the customer is asking for (or the intent of the conversation), as well as the amount of emotional energy (or sentiment), that’s hidden in the word usage of the email or chat conversation. This unlocks the ability to process the message with RPA using specific frequently used responses that can be tailored to the customer. Alternatively, the conversation could be routed to a person to help manage the interaction in a more personal way to improve the customer’s experience.
These are just a couple examples of combining RPA and another technology to support a richer experience. Another example could include automated evaluation of purchasing history, using AI to recognize patterns during the buying interactions. This could be used to improve marketing campaigns or influence the decision and ability to offer new, related goods and services. The automated data gathering potential that the RPA platform can feed into one or more AI engines can greatly increase the way that business is optimized both internally, by reviewing historical processing of transactions, and externally, by pulling data from suppliers and logistics organizations to improve supply-chain management.
Intelligent Automation is a Powerful Combination of RPA and AI
In addition to supporting decisions that the thinking part of intelligent automation offers, the doing is a critical component that cannot be overlooked. RPA and AI together offer a compelling combination of extracting information from various corporate systems and evaluating that data using powerful algorithms. But the real magic comes from acting on the decisions that were made by the AI engine. On its own, AI is like a disconnected brain. Without the capabilities offered by RPA, AI would require special connections to database systems to pull in the data it needs to evaluate, as well as coding for application programming interfaces (API) so that the AI decisions can yield an action.
For example, in the case of a banking system that is evaluating transactions for fraud, data is pulled from accounts to be fed into an AI algorithm. The AI engine evaluates whether the transactions match the pattern with which the customer typically spends their money, and whether that customer is breaking any laws by funding illegal activities. But simply identifying these activities is not enough.
If a transaction is found to be fraudulent or, at least, deserving of additional inspection, the AI engine would need to be able to act by stopping the transaction and alerting someone about potential fraud. Those actions might be complex, such as alerting the government, freezing the transaction, communicating with internal examiners, and sending a notice to the customer.
With an RPA system, those actions can be handled quickly and easily by interacting with the various systems inside and outside of the organization, without requiring programming. Without the doing aspect that RPA can offer, the thinking component offered by AI requires a lot more effort. Therefore, intelligent automation is a powerful combination of RPA and AI.
A term defined by Gartner, “hyperautomation” takes the concept of intelligent automation and extends it to include additional applications. RPA with business process management (BPM) tools and AI and analytics are combined to create a workflow processing framework that directs decision-making and analysis towards AI algorithms while capturing data on the operations of the business. This type of framework lends itself to a more complex interaction of various AI technologies like OCR, IDP, and NLP as they interact with data from the company and metadata about the way the company routes work and processes information.
The general goal is to create a digital twin organization (DTO) that can approximate the operations of the company in a way that can support additional modeling to improve the operations. For example, a DTO might capture data about their pricing and sales to model the impact of managing profitability in specific markets. The digital-twin model, based on prior historical data, could model the impact of pricing changes on logistics, warehousing costs, personnel costs in the shipping departments, supply-chain change requirements, and overall profitability. These changes then could be modeled in real-time across the entire company even with limited or selective market pricing changes applied.
The automated feedback from the systems could then make the adjustments necessary on personnel staffing for various shifts and apply those changes to multiple systems while tracking everything in real-time. Again, this is just one example of the hyperautomation framework. It takes intelligent automation beyond simply applying digital smarts to an automated process. It opens up the possibility of adjusting parameters in the business based on feedback received from the changes and the decisions from the AI algorithms.
Automation and the Future of Work
With the advances in automation technologies, clearly RPA, intelligent automation, and hyperautomation will continue to evolve as better AI engines support decision-making within the operations of a company. Starting out may seem daunting but thinking big will often pay big dividends as your organization builds the layers of automation from RPA through hyperautomation. The biggest risk is not thinking big enough while competitors embrace and adopt these technologies to gain a competitive advantage.