June 8, 2021
Intelligent process automation vs robotic process automation
RPA Business Analyst
When it comes to discussing the "intelligent process automation (IPA) vs robotic process automation (RPA)" issue, the choice of terms is highly controversial and can cause enormous confusion. That's because RPA is a multifaceted ensemble of business process automation technologies that may manifest in many possible configurations. One of these is IPA.
The logical conclusion of this assumption is that making a direct performance comparison between RPA (as a wide set of automation solutions) and IPA (as a manifestation of the former) would be like asking what's better between pasta and spaghetti.
Such a clear differentiation and opposition between the two terms may be considered as a reminiscence of past years when traditional RPA solutions were typically limited to deploying rule-based bots alone.
This is no longer the case because the involvement of AI in RPA tools is not an exception but an expanding trend. Very soon, IPA and RPA might be two almost fully overlapping terms, as AI-powered bots could become the new standard: according to Gartner, by 2022, 80% of organizations that deployed RPA will introduce AI too.
RPA as a whole has done for the service sector what the widespread deployment of industrial robots had made possible in factories a century earlier. RPA software robots, programmed to mimic the actions taken by physical workers when fulfilling their duties, were delegated a wide range of repetitive, time-consuming and tedious tasks once performed manually by human operators. Among RPA use cases, we may include entering customer information in corporate databases, filling in forms and reports, recording transactions or other sensitive data for accounting purposes, and so on.
The greatest result achieved by the companies investing in RPA software development and deploying this technology has been a significant boost in terms of productivity. According to Protiviti's 2019 Global RPA Survey, increasing productivity ranks first among all major benefits of adopting RPA, regardless of industry and implementation maturity stage.
Specifically, the percentage of respondents reporting moderate, large, or very large increases varies from 50% in the tech-media-telecommunications sectors to 23% in energy utilities during the previous year.
RPA adoption also enables a significant number of companies to reduce operating costs (up to 54% of the tech-media-telecom corporations and 40% of the financial organizations surveyed by Protiviti). Furthermore, bots greatly increase the accuracy of business process execution while allowing employees to focus on more motivating activities involving data-driven decision-making, analytics, and public relations.
Other noteworthy benefits relate to auditing and tax compliance. The introduction of bots and the consequent reduction of human interactions with sensitive data not only significantly reduces the risk of disastrous errors that could end in lawsuits or fines but also prevents the occurrence of fraud.
Finally, RPA opens the door to digital transformation with reasonably low investments. Software robots can fit into legacy systems and automate certain business processes without the need for costly back-end integrations and without modifying legacy infrastructures.
Traditional rule-based robots, whose adoption was predominant in previous generations of RPA solutions, are high-performing yet a bit conservative. They shine when it comes to managing repetitive, rule-based business tasks and processing well-structured digital data or standard input and output formats, such as numbers on a spreadsheet.
We may think of rule-based RPA bots as hobbits from The Lord of the Rings. Our little country villagers excel at doing repetitive manual tasks like gardening or farming. If they had a laptop, they would spend whole days filling out forms and dealing with reports or invoices.
Such bots’ talent in handling paperwork can be harnessed in virtually any industry. For example, the deployment of RPA in HR streamlines the review of candidates' resumes, thanks to the ability of robots to recognize and collect salient information such as years of experience. At the same time, RPA in finance is leveraged to extract data from supplier invoices and reorganize it into standard, easy-to-consult forms, or to self-fill and mail invoices to customers.
What basic robots can't do so well, however, is autonomously deal with rule exceptions or understand the context. On top of that, these bots cannot learn or improve from the information they process because they just handle it without actually understanding its contents.
This is where intelligent automation technologies come into play to make bots truly aware of their operating environment while also assisting with their self-learning.
IPA is a "sparkling" combination of RPA tools, machine learning, and cognitive technologies such as computer vision and natural language processing.
According to Deloitte's 2019 Automation with Intelligence report, the most popular AI solutions involved in intelligent automation strategies were machine learning and predictive algorithms (already implemented by 48% of the respondents), even if deep learning may become the most prominent in the future.
ML arguably plays the main role due to its ability to unlock a bunch of valuable cognitive skills to augment RPA robots with, such as:
One of the most fascinating features of ML algorithms is that they are able to recognize specific patterns among the data they are "fed" with and build mathematical models representing these relationships. Then, they can leverage these models to make predictions or decisions without even being explicitly programmed to do so.
The more data they can analyze, the more these models are refined through an increasing number of relationships and patterns identified. In this way, algorithms sharpen their performance with a real and almost human learning process.
Let’s consider an algorithm created to recognize different cat breeds in an image. Its purpose was certainly established by human programmers, but the path to achieving it is learned by the machine itself after proper data training. The computer can identify some common traits among cat images and some recurring differences in shape, size, and color (such as the black spot on the snout of Siamese cats) to distinguish between breeds.
Assuming we don't really want to compare cats to employees but just offer easy-to-understand examples, something similar can also be done when it comes to enhancing robots deployed in business processes.
In the field of human resources, for example, ML-powered bots can be trained with data about candidates' and employees’ performance, skills, know-how, and achievements. Algorithms will recognize patterns and classify professionals into different archetypes with similar traits.
They may even understand which types of workers tend to excel or struggle in performing some tasks and offer useful insights during the selection and training processes. This turns traditional automation into intelligent automation and allows robots to follow an analytical, data-driven approach instead of a rigid, rule-based logic.
It should be clear by now how enhancing bots with AI represents a great leap forward on the automation journey. However, it may be useful to clarify how intelligent bots can perform in particular business use cases.
Having many different file and format types used by each vendor can be challenging to interpret. In these cases, IPA bots can contribute with higher extraction capabilities, collecting salient data from each invoice and reprocessing them in predefined templates ready to be sent to the competent offices.
Another situation in which IPA bots' exceptional data extraction capability proves invaluable is the management of insurance claims. Where basic RPA robots could automate the acquisition of written data from an accident report, IPA robots can pull out unstructured data from driver license scans or car damage photos, which are quite common attachments in similar contexts.
In addition to classifying incoming emails based on rules such as the presence of keywords in the text or the type of attachment, IPA robots can study the behavior of staff who usually receive emails, identify repeated actions (such as sharing it with other colleagues in the same office or putting it into the spam folder), and learn to do the same automatically.
IPA robots can even understand the path of standard messages within the company and route any email containing an invoice or purchase order to the accountant responsible for approval.
What if we had to automate the sending of outgoing emails instead of sorting the incoming ones? Again, bots can help out.
IPA can leverage its enhanced analytical capabilities to enable greater personalization, taking into account customer data and preferences and tailoring the commercial offer with the highest purchase potential.
Forgetting to register your attendance at work is much more unlikely when a bot does it for you. Intelligent bots can check the log-in of employees when they access the company network or send them a reminder if they have forgotten to register their presence or absence in the appropriate forms. In a more advanced scenario, they can also detect the arrival of employees in their offices via corporate cameras and take care of recording their presence without the need for manual operations.
Considering what has been said so far, the answer to this question should be a resounding "yes". Such enthusiasm is clearly confirmed by very promising data.
Based on Deloitte's aforementioned report, companies adopting IPA increased their revenues by an average of 8.5%, compared to the 2.9% of corporations that chose to implement basic RPA solutions. Intelligent automation also drives improvements across customer experience, accuracy, and analytics:
Business executives also asserted that the best way to increase the competitiveness of their companies is the integration of RPA tools with AI, and estimated that intelligent automation would ensure an average cost reduction of 22% and an increase in revenues of 11% in the following three years.
Obviously, these achievements won't come for free; however, market-leading RPA platforms, UiPath included, already integrate a range of AI capabilities into their products out of the box. This should facilitate the onboarding of intelligent technologies by enterprises, along with opening up more advanced process automation opportunities.
Learn about the essential role of RPA in insurance and some of its key use cases.
RPA in real estate is a valuable ally but still far from its maximum potential. Find out how to leverage it to the fullest.
Discover the contribution of RPA to finance and accounting, its top use cases, and the good practices to implement it properly.
Learn how the application of RPA in healthcare can alleviate the administrative burden and increase operational efficiencies.
Discover the current state of RPA in HR management and its main use cases.
Learn about the most relevant RPA use cases by industry and some limitations and requirements to take into account.
Learn the most potent use cases for RPA in banking and figure out the best implementation strategies with Itransition.
Discover the most potent use cases of RPA in the automotive sector and learn the best implementation strategies.