A Definitive Guide to RPA Terminology

HPA News

RPA terminology

RPA, IPA, DPA, NLP, Hyperautomation—The RPA industry has more than its fair share of terminology (and frankly, far too many acronyms!) In this post, we translate the vast RPA lexicon to get automation novices up to speed quickly or serve as a refresher for those who are well on their way with their RPA journey. Of course, if we missed a term you think would be helpful for others, please drop us a comment below.

API: Application Programming Interface, or API, allow applications to interact with data from external sources. APIs act as a bridge between two or more existing applications, allowing a developer to define how they communicate with each other. APIs allow for a greater degree of connectivity throughout programs, and user-interface (See UI below) based automation can be combined with APIs for even greater capability and speed. Learn more about how APIs work with UI-based automation.

Artificial Intelligence (AI): AI is a broad category of computer science with many branches, including machine learning, deep learning, natural language processing, and other advanced capabilities. AI enables machines to mimic human decision-making, identify patterns in data, and learn and adjust to new inputs. AI models use a combination of mathematics, computer science, linguistics, psychology, and statistics to understand and utilize both structured and unstructured data. Common applications of AI in RPA include computer-aided translation like fuzzy logic, path finding, string and pattern matching, packing, computer vision and graph interpretation, outlier detection, and anomalous notification.

Attended automation: Attended automation, sometimes referred to as Remote Desktop Automation (RDA), is deployed to user desktops and robots are configured to work alongside human workers. Attended robots can gather data from multiple sources instantaneously and aide human workers in better decision-making. Humans and robots can also pass work off to each other using prescribed triggers. For example, a virtual assistant robot can be programmed to handle certain queries independently, and alert a human for resolution when the queries surpass its capabilities.

Business Process Management (BPM): BPM is the practice of continually evaluating business processes for the purpose of making processes more efficient and ensuring process outputs are align with company goals. BPM coordinates well with RPA as processes can be optimized before being automated to streamline implementation and achieve greater stability.

Center of Excellence (COE): The Center of Excellence acts as a steering committee for the RPA initiative, guiding the program’s design, development, and governance. The COE is comprised of business users, process owners, business analysts, and programmers to ensure every aspect of the automation initiative is properly supported. Selecting RPA software, selecting processes for automation, determining the business case, securing buy-in from the C Suite, coordinating with departments across the enterprise, and tracking the program’s overall success are all functions of the RPA COE. Learn why a COE is so important to your success.

Computer vision: Computer vision is a subset of artificial intelligence (AI) where computers are trained to interpret and understand images and visual stimuli. Specific to RPA, robots can be trained to identify objects within dynamic interfaces or across virtual desktops interfaces (VDIs). Examples of dynamic interfaces include Citrix, VMWare, and Microsoft RDP.

Deep learning: Deep learning is a subset of AI and part of a broader family of machine learning methods. With deep learning, a computer model can learn to perform classification tasks directly from image, text, or sound. Deep learning models are trained using artificial neural networks, which were inspired by the dynamic information processing capabilities of the human brain. The human brain can ingest and make sense of a wide variety of information in a wide variety of formats. Whereas humans can use their brains, eyes, ears, touch, and taste to make sense of the world around them, computers must ingest even larger datasets to make those same distinctions and categorizations. As a simple example, to identify an image of an elephant that is surrounded by text on a piece of paper, the computer must first identify the edge of the paper, distinguish text from image, and distinguish elephant from all other animals. Such a simple process for humans is incredibly complex for computers.

Digital transformation: Digital transformation is the process of using digital technologies to modernize the business as a whole. Digital transformation initiatives typically entail sunsetting legacy technologies and inefficient business processes to adopt a more synergistic operating model that is unified by new digital technologies.

Hyperautomation: Hyperautomation is the combined utilization of advanced capabilities like AI, machine learning, natural language processing (NLP), optical character recognition (OCR), and process mining to augment human intelligence and achieve end-to-end automation of business processes.

Infrastructure: Robots, like wild horses, need space to run. RPA software is typically deployed to virtual machines, either on-premises or to a hybrid cloud, virtual private cloud, or public cloud. Cloud deployments can be hosted in environments like Amazon AWS, Microsoft Azure, and Google Cloud Platform.

Intelligent automation (IA): Intelligent automation, also known as intelligent process automation (IPA), is a term used to describe the combination of RPA and advanced capabilities like AI, NLP, and OCR.

Machine Learning (ML): ML is a subset of artificial intelligence (AI) being applied in the RPA industry today. Machine learning is the application of AI to provide models the ability to learn and make decisions without being explicitly programmed to do so. In RPA, ML can advance robots beyond rote process execution and allow them to take on tasks that traditionally required some degree of human decision-making. Learn more about machine learning in RPA.

Natural language processing (NLP): NLP gives computers the ability to read, understand, and derive meaning from human languages. A common example of NLP is chatbots, which read or listen to human inputs to determine their intent and route accordingly. NLP can also be used to parse unstructured data like ‘contact us’ forms, emails, and voicemails.

Optical Character Recognition (OCR): OCR gives computers the ability to identify and extract information from images. For example, OCR can be used to extract data from scanned bank statements, pay stubs or W2s to be used by robots processing mortgage loans.

Proof of Concept (POC): In the world of RPA, a POC is typically conducted at the beginning of the RPA initiative to prove out the automation business case, set the expectation for ROI it could generate, and aide in selection of the best RPA software for the business.

Return on Investment (ROI): Time and cost savings are the base metrics for determining RPA’s ROI. Other factors like overhead cost avoidance, penalty avoidance, cycle time improvements, and employee or customer satisfaction are equally as impactful, but much more difficult to measure. Time and cost ROI is determined by comparing the time and cost of manual processing vs. the overall cost of licensing and personnel to support the program. To determine manual processing cost, calculate how long it takes to complete the process manually (referred to as manual handling time) and the cost of the employee per that amount of time. Employee pay typically includes base salary plus benefits, PTO, and taxes, which is usually 1.25 to 1.4 times the base salary.

Robotic Process Automation (RPA): RPA is the utilization of robots to perform repetitive, rules-based tasks that utilized structured data. The robot is programmed to replicate and execute against the specific steps taken by a human to perform a business process, within the same applications employees use.

RPA-as-a-Service (RPAaaS): RPAaaS is a consumption model similar to the SaaS model, where the client pays for the use of the technology versus purchasing the software outright or via license. RPAaaS solutions allow companies to access automation without taking on the risk and resource constraints associated with building an RPA solution in-house. Benefits include the elimination of complex licensing, instant availability of the latest version of the technology, and access to RPA expertise to guarantee success, in addition to the risk and resource constraints. RPAaaS can be more cost-effective than licensed RPA. Learn more about RPAaaS here.

Structured data: Structured data is comprised of information that is predictable and repetitive, like names and addresses in a standardized entry form. RPA robots can easily recognize data with a pattern, like numbers that are in the same section of every form. Structured data is highly organized and formatted so it’s easily searchable in databases. RPA robots work best with structured data, since it is so straightforward and ready for use in automation.

TCO: Total cost of ownership (TCO) refers to how much it will cost your business to own and operate your RPA software. Licensing costs only represent about 25-30% of the total costs associated with RPA. The remaining 70-75% represents the cost of support personnel, the underrepresented category of cost that isn’t discussed nearly as much as it should be. TCO includes one-time costs, like establishing a CoE, and ongoing costs, like annual overhead for support staff, ongoing training, and continuous bot monitoring. Learn how to calculate TCO.

UI-based automation: With User interface (UI) automation, the robots interact with the interface of the application that is being utilized in the automated process. Robots login just as an employee would and replicate the same steps they take to perform the process manually. UI-based automation can be combined with APIs to create automated scripts that contain fewer steps, are less prone to breakage, and have a shorter runtime.

Unattended automation: Unattended robots run independently on virtual machines and do not require inputs from humans. They execute business processes utilizing a prescribed set of instructions from which they do not deviate. Unattended robots are ideal for rules-based, repetitive tasks that utilize structured data and stable applications.

Unstructured data: Unstructured text, video, audio files, e-mails, image-based PDFs and satellite imagery are some examples of unstructured or semi-structured data. RPA robots cannot interact with unstructured data, since they need labelled and defined fields to work with. ML models are trained to work with other cognitive capabilities to extract and structure unstructured data for automation. OCR engines can be used to identify, extract, and categorize data from scanned images and PDFs. NLP can be trained to understand sentiment in free-form text, like customer service emails, chats, and voice inputs.

Virtual workforce: A flexible, self-managing RPA platform is known as a virtual workforce. While it is not considered ‘intelligent’ in the sense that it can learn or make decisions, a virtual workforce is highly adaptable and can multitask. Virtual workforces can support AI and business processes, providing high levels of security, consistency, and predictability.