So Long Automation, Hello Hyperautomation
As many organizations look to automation to drive productivity, automation itself is evolving and changing as these same organizations start implementing it. The growing interest in the technology was underlined by the recent UiPath IPO that gave the company an initial market capitalization of $29.1 billion.
For organizations that have embraced task-driven automation technologies such as robotic process automation (RPA), it is starting to become clear that a single, task-based automation tool is not enough to implement efficient, resilient digital business processes.
Instead, according to Gartner, they are now turning to hyperautomation, a process whereby organizations automate whatever can be automated and transition from RPA used for simple, rule-based task automation with structured data to RPA combined with AI that can automate complex processes with unstructured data and a significant level of ambiguity. The data generated from intelligent automation can then be interpreted by humans to make business decisions.
Behind it all is data. With the exponential growth in the volume of enterprise data, companies need to adopt new strategies like hyperautomation to keep pace with their customer and employee experiences, said Kurt Trauth, senior vice president of customer experience strategy and analytics at Charlotte, N.C.-based Stratifyd, a developer of AI and analytics platforms.
Hyperautomation Feeds on Data
One of the critical elements of hyperautomation, Trauth said, is the way it will democratize data and advanced analytics tools like AI-powered speech and text analytics to more business users, rather than siloing away these capabilities in the hands of IT teams and data scientists.
In practice, hyperautomation provides business leaders the flexibility to make data-driven decisions in near real time. In particular, the adoption of no-code artificial intelligence, automation and text analytics will give business leaders the power to unlock insights from the 80% of unstructured customer interaction data they are missing out on today.
“If business leaders can take the burden of manual data preparation and analytics off their teams with AI and automation, they can focus more of their time on responding to and modifying the customer experience,” he said.
"For example, customer experience teams can use hyperautomation to monitor call center interactions to predict [Net Promoter Scores] without the typical lag time in surveys, or they can uncover real-time insights from chatbots to highlight improvements for the digital experience team."
Expanding the Automated Business Process Universe
Hyperautomation is a response to the limits of RPA and to code in general. The first thing to understand is that, for a business process to be automated, it must be deterministic. That is, you must know exactly what to do in response to a given input, said Chris Nicholson, CEO of San Francisco-based Pathmind, a company that applies AI to industrial operations.
There is no room for ambiguity. Code is deterministic and operates on the principle of "if this, then that," never "if it is kind of this, then maybe that or this other thing." But many business tasks are not deterministic. There's wiggle room and ambiguity. Incoming data can usually be interpreted in several ways and there might be more than one way to deal with it. That does not work with code or with RPA.
"Humans are underrated," Nicholson said. "We know how to respond to new events that we were not specifically programmed for. We often know how to handle corner cases and exceptions, how to bend the rules when necessary to accomplish a larger goal. Software and RPA can’t do that."
Hyperautomation can help by enabling humans to make decisions more efficiently even amid ambiguity. This is called decision support. As we move away from the deterministic world of RPA and structured data, we get into unstructured data like an image or a long document on a PDF. When you are analyzing text in natural language, there's always ambiguity.
“The only writings we produce that are not ambiguous are code and mathematical proofs." Nicholson said. "So to respond to the ambiguity of natural language, you need probabilistic models. A probabilistic model is one that says: The data you have given me 'probably' means X, and therefore you should probably do Y."
Feeding those kinds of predictions from a probabilistic machine learning model to a human who is familiar with the situation can vastly accelerate work because it points them in the right direction.
“That's one thing that people mean by hyperautomation. Humans apply context awareness and ambiguity handling to business processes, which code alone cannot do. But trained AI agents can help people do that faster,” Nicholson said.
Szükség esetén segítséget nyújtunk
Felkeltettük érdeklődését, és kérdései lennének? További információkért keressen minket weboldalunkon szereplő elérhetőségeinken. Keressen minket bizalommal!