Cognitive Computing: Real-world applications
for an emerging technology
IBM began its Watson research project nearly a decade ago, and the resulting cognitive computing system competed in and won a high-profile Jeopardy contest in 2011. The intent was to successfully build a system that proved a computer can learn, think and understand like a human. Watson combined natural language processing, machine learning and knowledge representation in a way that no other systems had in the past. Watson ingested the questions, searched its repository for information, developed and analyzed hypotheses, and produced answers that were also in natural language form.
The combination of those capabilities is important; their integration is equally so. “It’s not just the pieces and technology but the fact that they are so well integrated that they can influence each other,” says Susan Feldman, CEO of Synthexis. (Please see related article by Susan Feldman on Page 1, KMWorld, Vol. 24, Issue 8, What makes a computer system cognitive?.) The orchestration is a vital part of the technology. This is a new kind of computing that is aimed at very complex problems, and its ability to develop meaningful conclusions from diverse sources is essential.”
Over the past few years, IBM worked with partners to implement the technology for healthcare, financial services and horizontal (cross-industry) applications. In 2014, the company announced a major commitment to establish a dedicated Watson business unit. It also opened up Watson as a developer platform, offering technical support and allocating funding for venture investments to seed businesses building Watson-powered apps or incorporating Watson capabilities into existing products.
Extra layer of intelligence
One of the first customers to begin incorporating IBM Watson into its product was Welltok. The company was founded prior to the commercialization of IBM Watson; its CaféWell platform guides and encourages consumers to engage in healthy behaviors that lead to optimal health. Sponsored by healthcare insurers and providers, Welltok was also the first company to receive support from IBM’s Watson fund. Consumers are connected with the resources and programs that are most relevant to them based on health status, available benefits, interests, demographics and other factors.
“We saw the opportunity to add a layer of intelligence and other capabilities to the application,” says co-founder Jeff Cohen. “Adding cognitive computing enables us to provide even more relevant recommendations to consumers, to connect with them at a more personal level and to modify those recommendations as the system learns more about the person.” The addition of IBM Watson into CaféWell created the CaféWell Concierge application, which lets consumers use natural language to interact with it, get recommendations on a variety of topics ranging from condition management to healthy dining options and benefit information, and allows Welltok to provide a personal health concierge for every eligible consumer.
Commercialization is taking place across a wide range of industries and a growing number of countries, according to Alexa Swainson Barreveld, VP, Watson products & solutions. “Cognitive computing is filling a tremendous need,” she says. “People want to solve complex problems around unstructured information, and they have not been able to do that before.” The Watson Oncology Advisor developed by the Memorial Sloan Kettering Cancer Center provides a model of the way cognitive computing can be used. “We want to enable a partnership between people and the machines they work with, so each can use their strengths,” she says.
User guidance
Applications based on Watson are being developed for both the vertical and the horizontal markets. CognitiveScale offers four different applications through its cloud product, Insight Fabric. The healthcare application provides guided care for chronic diseases to help healthcare providers better manage those conditions. The guided commerce application, designed for retail sales, uses cognitive computing to drive key metrics and personalize offerings to drive conversion. Guided service for HR is a horizontal product designed to increase employee productivity by expediting such business processes as expense reporting and help desk responses. Guided procurement provides insights that improve customer services.
“We picked those four applications because they all have a high ROI potential,” says Matt Sanchez, founder and CTO of CognitiveScale. “These are first-of-a-kind products but not one-of-a-kind, so they can be replicated in many organizations.”
In one large consumer packaged goods company, the guided service application was geared to support the SAP software used by the company. “When an employee runs into a problem, the usual approach was to create a trouble ticket, which can cost from $24 to $160 and can take days to resolve,” Sanchez explains. After doing a pilot project, the company was able to achieve a 30 percent reduction in tickets. “When a customer has a million tickets a year, the savings are huge,” Sanchez says.
Engaging customers
Although IBM is the leader in the cognitive computing field, the industry is growing at a rapid rate and there is room for new entrants. CustomerMatrix was founded two years ago to focus on producing new revenue in sales and customer engagement by providing real-time information to client managers and customer service representatives. “We see cognitive computing as a truly disruptive technology,” says Guy Mounier, CEO and co-founder. “Previously, companies would spend a fortune putting together a data warehouse, but account managers and service representatives would be at a loss when facing a customer in real time. Cognitive computing synthesizes information to produce revenue generating recommendations proactively in real time.”
The CustomerMatrix platform consists of several components that collect, process and enrich information from multiple internal and external sources; structure and analyze it to find patterns and relationships; seek out and incorporate missing information; create user profiles to determine what information is needed; and send out ActionAlert suggestions, ranked by revenue impact value. Designed to be embedded in user workflows, such as records management portals or customer relationship management (CRM) systems, CustomerMatrix uses a combination of federated search, advanced semantic analytics and cognitive analytics to create the metadata hub that provides the basis for understanding and recommendations.
CustomerMatrix has found a lot of traction in the financial services industry. “In this industry, products tend to be commodities,” says Mounier. “The best way to differentiate is to capitalize on the customer relationship to create an edge.” Because it can look at so much data, cognitive computing can pick up on subtleties that traditional analytics would miss. “What is exciting to our customers is that cognitive computing provides a transformative impact on revenue, not a marginal one,” Mounier continues. “One client increased its upsell by a factor of 10.” With numbers like that, even customer care can become a profit center rather than a cost center.