Every serious Center of Automation needs two core foundational technologies: RPA and APA. RPA stands for Robotic Process Automation, and APA stands for Analytical Process Automation. But what does that mean? Although both technologies perform automated tasks, they do so in different ways. Some processes are better suited for RPA-built solutions, while others align more effectively with an APA solution. Often, the need arises for both to efficiently automate a process.
By choosing the right technology for each process – in other words, the right tool for the job – companies can significantly reduce the effort required to build automations and maximize the utilization of their software. As a result, we witness a dramatic acceleration of time-to-value, increased ROI, and a substantial reduction in total cost of ownership. While all of this holds true, we encounter many Centers of Automation (CoAs) that might possess both sister technologies but employ them in separate groups. Alternatively, some may have just one tool and struggle to fit it into inappropriate use cases. Based on my hands-on experience assisting customers with this issue, my assumption is that most companies either regard RPA as a universal automation tool or remain unaware of the existence of APA. The aim of this post is to demystify the crucial importance of both tools and to help better comprehend when to utilize RPA, APA, or even both.
RPA and APA solutions process information using entirely different approaches. To make this easier to grasp, consider visualizing a spreadsheet, although this concept applies to various data sets such as databases, Excel sheets, CRM and ERP systems. In the example below, rows represent individual employee records, with each row containing information associated with one person. Columns hold specific data points. For instance, the first column contains the first names of individuals, while the first row encompasses first name, last name, role, address, email, and phone number – specifically for John Smith.
If my automation task requires processing each record sequentially, the best approach would involve an RPA platform. For instance, imagine these records are new employees who need to be onboarded. I would take one record, generate an account in the ERP system, assign a company email, and complete other necessary tasks in the onboarding process. This type of process is transactional in nature, necessitating single-threaded technologies like RPA to tackle it. In such cases, we need to handle business exceptions, errors, tight SLAs, and create associated audit trails individually – tasks effectively addressed by RPA solutions. Keep in mind that while the data source in this example is an Excel spreadsheet, the data source, tasks, and destinations are immaterial. The critical factor is how records are processed – one at a time.
A different context for the same example aids in understanding where an APA solution would be applicable. In the same Excel spreadsheet, let's assume the goal is to arrange the records in alphabetical order, remove leading or trailing white spaces, standardize addresses, and merge first and last names into a single cell. Subsequently, additional data needs to be added from another source. Imagine the second spreadsheet containing first and last names, along with corresponding zip codes. By matching names in both datasets, I aim to append the zip codes to the correct records in the first spreadsheet. Following this, I might want to email it, upload it to a system, or save it somewhere – all achievable using an APA solution. The way in which I process this information is all at once. Why is APA better suited for this? It is a multi-threaded solution, capable of applying these functions to all records concurrently. Visualize attempting the same with RPA. Just comparing two spreadsheets requires comparing the first row to the first row in the other source, then the first to the second, and so on. Even for something as small as two spreadsheets with ten rows each, you end up executing 100 actions one at a time with an RPA solution. APA, on the other hand, processes every record simultaneously, significantly reducing runtimes to milliseconds and simplifying the necessary programming. Moreover, as volumes increase, the contrast becomes more pronounced. Combining two data sources into one data stream with 1,000 rows each – still relatively modest for most customer datasets – would entail 1,000,000 actions. If RPA were used, runtimes would become extensive, consuming software time and resulting in backlogs of queued automations awaiting execution. APA's quicker processing and simpler programmability enable more efficient development and resource allocation across CoA platforms as volumes scale.
These technologies excel in their respective functions, and there's a time and place for both. However, combining them enhances functionality across platforms. An obvious scenario for utilizing both is when processes require both single-threaded and multi-threaded approaches. For example, continuing from the previous scenario: after merging the spreadsheets, I want to input the data into my CRM. However, I need to create each record individually in my system. Logging in, creating a contact, filling in the fields for each row, saving, and moving to the next row – an ideal task for RPA. In this case, I can seamlessly transition results from my APA process to my RPA process, streamlining development and conserving software resources.
This example also highlights another crucial role played by RPA: UI Automation and Computer Vision. RPA's ability to manipulate application user interfaces, interact with on-screen elements, type, move, and click the mouse enhances APA's capacity to interact with applications lacking APIs and extends to interacting with virtually any interface. RPA is often employed to feed data to an APA process or to input results from an APA process into a system when API connectivity is unavailable, such as with external websites and legacy systems.
This article aims to clarify the distinctions between these two sister technologies, underscore the significance of knowing where to apply them, and highlight their foundational role in any Center of Automation. Lydonia has collaborated extensively with various Centers of Automation to demystify the relationship between RPA and APA software, creating scalable, efficient, and robust Hyper Automation solutions.
Vinny LaRocca
Senior Director, Automation Consultancy
Vinny is an experienced engineer with 10 years of experience in automation and artificial intelligence. Guided by his ability to understand and conceptualize complex, interconnected systems, Vinny has assisted customers worldwide with starting, building, expanding, and maintaining their automation capabilities. With world class technologies and industry leading minds, his team continues to help customers realize their full automation potential through automation, analytics, and artificial intelligence.
Before his start with Lydonia, Vinny worked as a Propulsion Systems Engineer. Working for UTC Aerospace, he helped design an AI driven, embedded sensor platform to be applied to guided missile technology. Using telemetry data fed back from the sensor arrays, Vinny and his team were able to reduce failure rates and improve overall accuracy of guided propulsion systems company wide. Shortly after, Vinny led a team of engineers for Jarvis Products developing custom AI Machine Vision systems for automation applications. In 2020 he led this team to win Integrated System of the Year from Fanuc for his innovative and pragmatic approach to AI development and deployment. His methodologies around artificial intelligence and how it applies to data and automation have been paramount in guiding customers to quick ROIs and huge revenue gains.
