Earlier this month, software vendor Pegasystems announced new AI and robotic automation capabilities in its CRM systems.
The addition of Pega Workforce Intelligence to Pega Customer Service and Pega Sales Automation applications allows businesses to uncover hidden patterns of inefficiency among all the actions every day in any application.
These include unnecessary toggling between applications, misplaced employee focus on activities that produce sub-optimal results, poorly performing networks and applications, best practices that are known only to the highest performing staff, and excessive reporting and meeting demands that overload schedules with low-value work.
Workforce Intelligence combines AI and robotic automation to understand how sales and service agents work and how the apps and processes work for them, automate redundant processes, reveal best practices, and suggest new approaches.
Pega Workforce Intelligence capabilities will be available in Q2 this year within Pega Customer Service and Pega Sales Automation.
Pegasystems CTO and vice president of product marketing Don Schuerman told iTWire that the company was using AI and automation "to cut through a lot of the complexity" faced by sales and support staff.
"Pragmatic AI technology" that is meaningful to organisations was available today, he said, for example in the form of predictive learning and machine learning systems that could make predictions while an agent was talking to a customer, proposing the right offer, conversation or service for that customer at that time, regardless of the channel they had elected to use.
Local companies had successfully used such systems, he said, pointing to a 13% improvement in close rates at the Commonwealth Bank, and the 25,000 software robots within Telstra's contact centres that handle tasks such as cutting and pasting information between systems so agents can concentrate on the customer to whom they are talking.
The wholesale replacement of multiple systems might not be cost-effective, but automation could make them easier to use, he said. Robotic automation could cut through the rat's nest of complexity, and AI could identify the opportunities for doing so.
Schuerman advocates using machine learning for what it is good at — handling massive data sets in real time — and then pairing that with human experience. For example, an unfettered AI system given the goal of maximising the outcomes of up-selling and cross-selling opportunities would probably sell anything a customer would take. But businesses needed to apply rules – a financial institution might have a policy forbidding the sale of a product that exceeds the customer's risk profile, he explained.
Apart from sales and support interactions with customers, the technology could also be applied in areas such as risk management (eg, customer retention), routing (eg, identifying the best person for a particular task), and sales management (eg, obtaining more accurate predictions of the likelihood of an opportunity being closed and the prospective close date).
Pegasystems' technology reveals opportunities for automation but keeps humans in the loop before actually automating the processes. The system in effect says "I've noticed people are doing this a lot – would you like me to automate it because this would be the benefit," and then human judgement and accountability is brought to bear in deciding whether or not to act on that recommendation.
Part of the problem with AI in business is that deep learning has been effectively applied to questions that have a right answer. For example, a system might identify a photograph as being of a boy with a bike. But in business, "the answer isn't black and white". In the customer experience context, are you trying to optimise share of wallet, net promoter score, retention, the cost of servicing, or some other goal, Schuerman asked. If the goal is lifetime value, what time horizon is appropriate? And is the objective necessarily the same for all customers and all processes?
So Pegasystems is working to produce tools that can cope with this complexity and lack of a simple answer.
The goal, Schuerman said, was to augment people to make them better at their jobs, without having to translate business concepts into code, and keeping people in control of the system.