In this segment of “Office Optional with Larry English,” Larry talks about hyperautomation – an innovative combination of robotic process automation and artificial intelligence.
In 2019, a McKinsey report found that only 55 percent of organizations found success with their automation program. Robotic process automation (RPA) promised a big ROI, but for some companies, capturing that value proved elusive. Fast-forward to today, and the meshing of RPA with AI technology — hyperautomation, or the marriage of multiple automation capabilities — has opened a whole new world of automation.
Much of the pre-AI trouble with RPA comes down to the complexity or variability of seemingly simple tasks.
For example, say you want a bot to analyze invoices and send them to the right contacts at your organization. Seems simple enough, right? Think again. Before the bot can handle the task, you’d need to standardize every invoice, or you’d have to program the bot to know how to handle dozens of different document variations. Suddenly, this “simple” RPA initiative became headachingly complex and expensive.
AI contributes a layer of intelligence to RPA that wasn’t previously possible. RPA on its own is limited to straightforward, structured and objective tasks. AI expands the possibilities to include capabilities with nuanced, subjective or unstructured data. By combining RPA and AI tools, you can automate a bigger chunk of your processes. For example, hyperautomation can:
Accurately read and sort documents.
Organizations can train an AI model to accurately read and sort complex documents, eliminating the need to manually program bots to know how to handle varying document formats. AI can even interpret handwriting, so organizations no longer have to manually enter handwritten documents into a digital format.
For instance, my company, Centric Consulting, recently partnered with UiPath, the world’s largest RPA vendor, to help World Wide Technology (WWT) apply hyperautomation to better manage the company’s high volume of purchase orders. Those purchase orders come in numerous formats and languages.
Without a standard template, RPA alone could not “read” and sort the documents without a separate setup for every format. With AI, however, it was another story.
We trained an AI tool on over 100 purchase orders from one of WWT’s vendors. By combining RPA with AI, WWT has successfully automated a process that previously required significant back-end work. This has been a strategic part of WWT’s hyperautomation journey.
“WWT is creating the capabilities to harness the power of GenAI and RPA to transform its business,” says Dave O’Toole, senior director of portfolio and product management at WWT. “By applying GenAI and RPA to its software development, sales engagement, customer service and operations, WWT aims to achieve higher quality, efficiency and productivity and better customer satisfaction and loyalty.”
Interpret free-form text.
AI can read and understand free-form text, chats, emails and other unstructured data. Say you have a team of people tasked with reading and sorting customer emails. Now, you can use an AI-powered bot to read those emails and send them along to the right team within your organization, freeing up employees to work on higher value-add tasks. And because AI models continue to learn, the accuracy improves over time.
Analyze images.
With AI, an RPA tool can analyze images and flag discrepancies or errors. For example, a food and beverage customer of UiPath needed to ensure the accuracy of product labels. The customer used an AI-powered automation to inspect and analyze design drafts, note discrepancies, and notify the appropriate design agencies. Before, this process was error-prone and required lots of time from lots of employees to complete. Now, each inspection takes a few minutes and is more accurate, saving the company from reputational risk and regulatory fines.
Produce and analyze complex reports.
Large organizations spend countless hours pulling, formatting and producing reports. An AI-enabled RPA tool makes this tedious backend work unnecessary. To share another UiPath client example, a global oil and gas company processed millions of pages of production operation reports each year.
Because these reports lacked a uniform layout and data structure, employees were forced to extract data manually, a painstaking process. With the help of hyperautomation, the company saved its employees countless hours, freeing people up to focus on higher-value work and speeding up the company’s decision-making process.
Classify complex information.
An Australian mutual bank needed to modernize how it investigated living expenses for assessing loans — the manual process was too time-consuming and error-prone. The bank first implemented a bot that could classify about half of a loan applicant’s transactions.
This improved the fully manual process but still left a lot of hands-on work. By adding in a custom machine learning model, the bank can now automate about 90 percent of the work involved in generating living expense reports, improving the speed and accuracy of the loan application process.
Ready, Set, Automate
The examples above merely scratch the surface of the high-impact possibilities of hyperautomation.
RPA was already growing at an impressive rate before AI. According to Gartner, the RPA software market reached $2.8 billion in 2022, representing a 22.1 percent growth rate. For context, the global software market grew just over 11 percent in 2022. Now that more advanced technology is widely available to organizations of every size, RPA is poised to continue making strides.
If you’ve been disappointed by RPA before, know two things: You’re not alone, and now is the time to try again. It’s not difficult to begin combining RPA and AI models. A growing number of out-of-the-box options are out there. UiPath, for instance, has prepackaged functions, such as sentiment classification, receipt reading, and text translation, that you can add to your automation functions. Another option is “tunable” AI, pre-built models you can train and customize.
Companies that pass on an opportunity to automate processes will fall behind their competitors who tackle automation and capture the efficiency gains and improvements in the working experience for their people.
A scorecard is a valuable tool to gauge the suitability of a process for automation, even with limited information available. We made a template to help.
