July 27, 2016      

Will artificial intelligence and automated processes replace human analysts as well as employees in manufacturing and the service industry? Despite the fears of robotics and AI taking jobs, new tools could make the human role in business more valuable, said analytics and automation practitioners.

About half of business leaders in the U.S. and U.K. said they have embraced robotics, according to a survey conducted by Genfour and 3Gem Media Group. The survey also found that 94 percent of respondents believe increasing levels of automation are inevitable, and about a third have plans and expect the transition to be fast.

“These statistics show a very different picture from the negative portrayal of the fear surrounding robots and automation we see all too often,” said James Hall, CEO of Genfour. “Business leaders see automation as a way to increase productivity, reduce repetitive and mundane tasks, and create efficiencies in the back office.”

While job fears may be overblown, people should understand how automated processes can affect different industries and free up talent for other, more strategic roles.

“The application of smart machines … is entering into the customer-service and knowledge domains, which were thought of as more protected from automation,” Hall told Robotics Business Review. “There’s other, better stuff for people to do.”

Identifying ‘levers’ for action

“BeyondCore analyzes data combinations and finds useful information,” said Jen Underwood, a technical product marketing consultant with BeyondCore Inc. “With smart data discovery, we can find the metrics and drivers that matter.”

Machine learning and regression analysis can be used together to present findings of key components and “levers” that can influence sales — and profits, she said.

“BeyondCore’s product takes away some menial work from analysts. It helps them go through a tedious process,” Underwood said. “This enables subject-matter experts to use their expertise to manipulate and transform data.”

“For example, in insurance, a law could have changed, and the machine would say, ‘Don’t hire anyone in New York,’ but a subject-matter expert is still needed to parse that information,” she said.

BeyondCore’s patented algorithms helped First Interstate BancSystem Inc. analyze business intelligence (BI) data more quickly and objectively. BeyondCore looked at transactional data sets with as many as 2 million rows and 10 columns to identify how different fees would affect the behavior of business credit card holders.

“Now we’re able to use BeyondCore to centralize data analytics to our team, and encourage other business units to not make decisions based on gut feel, but to leverage a tool like BeyondCore to tell us what is statistically relevant or not,” said Josh Botnen, director of data management and analytics at First Interstate Bank in Billings, Mont.

Applying AI to industrial IoT

Clients of San Mateo, Calif.-based BeyondCore must first identify what they want to measure or analyze. This could be a business process, or in the case of manufacturing, data generated by the emerging industrial Internet of Things (IoT). Mobile robots and other sensors can track a component or a product through the entire production and distribution cycle.

“From there, you can seed many columns of data, and the system can say which ones are irrelevant and find the signals in the noise,” Underwood explained.

Instead of trying to read dozens or hundreds of columns in a spreadsheet, analysts can use BeyondCore’s technology to analyze across thousands of variables.

For instance, BeyondCore Automated Analyst helped a Fortune 500 insurance company analyze 52 million policies and 10,000 variable combinations in 30 minutes.

“Right now, we’re still seeing very early adopters of IoT. Optimizing production processes — those are the types of things we are finding,” Underwood said. “We’re not doing a lot of analysis for them yet.”

“In precision agriculture, there’s Microsoft’s ‘connected cow,'” she noted. “There’s not a lot of analytics automation yet, but our chief technology officer came from that background.”

Actionable information

For e-commerce and supply chains, responsiveness is crucial to staying competitive. BeyondCore’s technology works with customer relationship management (CRM) systems to help act on BI.

“Our predictive algorithms allow for a query and forecast on the CRM integration level,” Underwood told Robotics Business Review. “We’re one of the only vendors that could not just make a prediction but also recommend an action.”

“Our most popular use case is integration with Salesforce for lead scoring and customizing call-ins,” she said. “It can be cooked into the client’s CRM system to give information to a sales rep on a phone call with a contact.”

Genfour's Robotic Workforce Report infographic

Underwood cited a McKinsey & Co. case study that looked at populations of people at risk for chronic ad costly health problems. By analyzing more than 100 variables across 30 million patients, BeyondCore was able to quickly identify certain patients for whom care coordination might help reduce costs.

“Where you can improve quality and information, starting to understand caseloads and automated processes, and make subsequent changes, that’s a virtuous cycle,” Hall said.

In another example, a food processor was experiencing profit leakage and turned to BeyondCore. Its application looked at 300,000 rows of data and in minutes found a pattern pointing to one of the company’s facilities for storing fresh vegetables.

“With descriptive, prescriptive tools, customers can answer ‘How can I improve it?'” Underwood said.

Of course, data is useful only if a human can understand it. “Our explanations are empowered by natural language processing, making the leap to the next level,” she said.

Genfour’s clients include roadside assistance company RAC, the Discovery Channel, PricewaterhouseCoopers, and McCormick & Co. It helps them with automated processes for account registration, payment processing, and financial controls.

“In automation, you start by automating things you’ve got that you can automate, and then extend it to things you’d like to do with people to but can’t,” Hall said.

AI evolving in parallel

“This year, we’ll see a lot of visualization tools, but they’re not necessarily able to make prescriptive ‘what-if’ analysis for optimization,” Underwood said. “The industry is going into deep learning and cognitive analysis to deal with unstructured data sources and provide context. We’re making analytics more human-friendly.”

She distinguished BeyondCore’s work from that of IBM Watson as an example of simulationist and modeling behavior versus prescriptive analytics.

“It’s a whole brand and division of IBM … but it’s more of a dashboard solution,” Underwood said.

Watson, perhaps best known for demonstrating AI’s capabilities on the Jeopardy quiz show, is “more of a vertical solution taking in unstructured documents rather than making recommendations,” she said.

“Watson can only look at historical data, but BeyondCore can produce measurable metrics and ‘what-if’ analysis for sales calls,” Underwood said.

Business process automation and prescriptive analytics should not be mistaken for less-sophisticated productivity tools, noted Genfour’s Hall.

Automated processes for government and HR

The U.S. government has been an early adopter of analytics tools. Underwood cited the Internal Revenue Service, the Defense Department, and the Bureau of Alcohol, Tobacco, and Firearms as examples.

“We have a couple of government clients,” said Hall. “Where we look, we see lots of opportunity, but there are longer cycles to start working with them.”

According to Genfour’s survey, human resources departments are split on the potential value of automation. “The split may reflect the maturity of those HR organizations — how operationally focused they are,” Hall said.

Genfour CEO James Hall

“Robotic process automation is more incremental and less disruptive than offshore outsourcing,” says Genfour CEO James Hall.

“HR is often one of the less systematic groups in companies,” he added. “It’s people-focused, but those who adopt it [find their work] facilitated.”

“HR is a good use case because of a high population of people to interact with, multiple systems — recruitment, payroll, as systems of record, 401(k) benefits, health insurance,” Hall said.

Genfour has automated its own onboarding, Hall noted. “We’ve also just done it for one of the Big 4 staffing companies in the U.S.,” he said.

However, hurdles remain for automated processes and analytics. About half of respondents to another survey said that they believe that business processes beyond data entry can be effectively automated, found Redwood Software Inc. and SharedServicesLink.com Ltd.

‘Bring on the robots’; AI no threat to jobs

“Analytics in fact doesn’t threaten jobs; it allows people to use their minds for strategy rather than spending 60 percent of their time crunching numbers,” Underwood said.

“In digital business and experimental businesses, there are short windows of opportunity,” she said. “With AI, people can think quickly and be creative and strategic.”

As industrial IoT and robotics continue to spread, the ability to gather data and make decisions with help from AI will likely spread from chief financial officers and similar roles to people managing production processes in real time.

“We have bunting up on our walls saying, ‘Bring on the robots,'” said Genfour’s Hall. “Robotic process automation is more incremental and less disruptive than offshore outsourcing.”

“There are lots of cases where people are bound to keep work onshore,” he said. “With a robotics model, the cost savings is improved, especially without moving offshore first.”

“In the U.K. and U.S., we’ve looked at recreating the Mumbai cost base,” Hall explained. “We look at a process, see if it’s rules-based and can be put in a logical sequence. We might find that 40 percent is automatable.”

“If we can find having a certain number of people onshore at a price point that’s equivalent, we can improve on high-volume transactions with speed and control,” he said. “Where should you have those processes running? As close to the source of decision-making as possible.”

“Managers may feel threatened by automated processes because there’s a changed burden on the management layer and not just on the employees they manage,” Hall added. “Ultimately, they want more throughput with fewer people.”

“Millennials aren’t as accepting of rote activities, so who are you going to get to do the work?” he asked.

Top-down vs. bottom-up automated processes

In fact, the spread of automation and business process re-engineering could change how organizations pursue projects, said Hall.

“Current organizations are used to doing top-down projects, identifiying requirements, and conducting software implementations,” he said. “This works the other way around. By introducing automation thinking into the broader workforce so it can actively participate, there should be broader engagement than with a lean methodology.”

“Young organizations and those in fast-moving industries are doing this today,” Hall said, mentioning e-commerce and ride-sharing companies. “Others have difficulty, but they’ll have to adapt their business models to compete.”

More on AI and Automated Processes: