Robotic Process Automation (RPA), Artificial Intelligence (AI) and Machine Learning (ML) are three distinct but overlapping areas of technology. They get conflated, with people sometimes asking “Is RPA AI?” To be very clear, RPA is not AI but can be used to assist AI with simple tasks, as we’ll explain below. RPA technology, especially the top RPA tools, continue to progress and develop more advanced features, which can make it confusing when looking at similar technologies, like AI. It’s worth taking a moment to focus on the RPA vs AI discussion and delve into how the industry views the intelligent automation trend. Essentially, RPA relies on AI in order for its software robots to “think” in the tasks they perform. However, an RPA bot can be quite simple in AI terms, and it may not be learning either.
Whether using RPA with AI or some other intelligent automation tool, the benefits of RPA are tremendous when understanding its strengths and limitations. When its strengths align with your business needs is when you'll maximize the ROI of RPA and realize the most savings.
Difference between RPA and AI
The human brain provides a good place to start when thinking about the difference between RPA and AI. The brain is capable of thinking and analysis. It can ingest data and spot patterns. Performing a task may require thinking and analysis, but the act of thinking and the specifics of a given task are not the same thing. One is general. The other is specific. That’s how to contrast AI with RPA. AI is about all the different ways a computer can analyze information and mimic the human mind’s capacity for critical judgment. RPA is about getting the computer to perform a task that requires some aspects of those thinking skills.
For example, an AI algorithm could see a thousand emails and “notice” that messages beginning with the word “Dear” are more likely to ask for a refund than messages that start with “Hello.” That’s pattern analysis. An RPA bot working from this AI analysis could be programmed to route “Hello” messages to one person in the office and “Dear” emails elsewhere.
Using RPA with AI and ML
IT Central Station members are using RPA with AI and ML. For instance, Shripad M., a Director of Operations at XLNC Technologies, a consultancy, has set up an Automation Anywhere IQ Bot to extract data from invoices. This cognitive bot then applies machine learning to determine if the data is not sitting in the correct columns. The bot puts the data back to the right columns. Oleskii D., a Team Leader at a large tech services company, similarly employs WorkFusion in document recognition tasks. As he explained, “It uses OCR [optical character recognition] and Machine Learning and works with digitized documents. It collects the fields inside the documents, based on the models that we trained with.”
Oleski’s last comment is particularly significant for understanding how RPA, AI and ML work together. The software enables his team to set up a model that can train the bot to perform the task. Over time, the model and training get better as the ML software learns from experience and improves itself.
RPA and AI and ML Use Cases
RPA use cases are vast and users are constantly discovering new applications, but here we'll focus on using RPA with AI and ML.
For Ravi K., a Program Manager for RPA who uses Automation Anywhere at a US-based real estate mortgage company, AI and ML are key drivers of his workload. He explained, “We do not want to train our bot constantly. We want to have something which has a self-learning bot.” In his use case, ML and AI combine to enable the bot to teach itself how to do its job better over time—without anyone having to spend time on training it.
Bharani K., an RPA Program Manager at Agility, a logistics company with over 10,000 employees, has a use case for Automation Anywhere that involves providing visibility of business-level and operational-level metrics. “It is extracting information, partial information, and real information,” he said. He values the software’s IQ Bot and Bot Insights because they enable his developer to use artificial intelligence to instruct the bots to build dashboards for the metrics.
Linking legacy Oracle systems is the UiPath use case for an IT Consultant at a large tech services company. The old systems had little connectivity through other integration methods. In the future, as he noted, “We are on our way to being able to link UiPath with machine learning and artificial intelligence to automate expenses.”
It is still early in the lifecycle for solutions that combine RPA, AI and ML. As the use cases reveal, the technologies are synergistic, if independent from one another. AI and ML make RPA better at RPA. With AI and ML, RPA bots can do more and even teach themselves to get better at their tasks over time. It will definitely be interesting to see how inventive RPA users put AI and ML to work as the three technologies mature together.