Next-Best-Action Designer for AI presents business users with a set of wizards they can use to automate processes infused with AI all on their own.
One of the biggest challenges when it comes to implementing artificial intelligence (AI) is getting the business leaders that own a specific process to concur with data science teams on how that process should be automated. All too often, data science teams don’t have enough visibility and insight into how a process really works to optimally apply machine and deep learning algorithms.
See also: Rob Thomas of IBM Talks AI Anywhere
To solve that problem Pegasystems has created a Next-Best-Action Designer for AI that presents business users with a set of wizards they can employ to automate processes infused with AI all on their own, says Matt Nolan, senior director for decision sciences at Pegasystems.
Next-Best-Action Designer for AI is an extension of the Pega Customer Decision Hub, which captures data during any customer engagement in real time that is then used to automatically apply business rules that dictate what type of customer experience will be surfaced to different classes of customers. A customer with a lower credit score, for example, would be presented with one set of options, while another customer would be presented with a wider variety of pre-approved service options.
AI, in theory, can be applied to automate those processes. However, business leaders tend to have much better insight in terms of where AI and predictive analytics can be applied to optimize customer experiences. Next-Best-Action Designer for AI allows business users to design those customer experiences by first defining their end goal and then relying on the tool to identify best practices to avoid missteps such as focusing on segmentation rather than one-to-one marketing when the overarching is to increase revenue per customer, says Nolan.
Other issues the software might surface include not using common strategies for both inbound and outbound targeting, or inadvertently mixing eligibility rules with propensity models to skew customer relevance scores.
If organizations don’t get business users to buy into how AI is to be applied most data science projects will die a quiet death on the vine, says Nolan. In fact, one of the reasons so many AI projects never make it past the incubation stage is the lack of input from business users, adds Nolan.
At the same time, however, Nolan cautions business users from trying to do too much with AI all at once. Instead, business users should focus on one specific channel at a time. The first channel usually takes about 90 days to optimize, with the amount of time required for the next channel declining with each successive engagement. Over time, organizations can then start to connect all those channels, advises Nolan.
“Then they’ll have one real-time console for controlling customer experiences,” says Nolan.
In effect, Pegasystems is moving to democratize AI within the context of an omni-channel approach to optimizing customer experiences. It may take a while for most organizations to achieve that goal. However, as data science technologies and techniques are embedded into applications, the days when organizations had to wait on data science teams to define the right set of algorithms to optimize a process may be coming to end sooner than anyone imagined.