Conversational AI for healthcare
We implemented an AI-powered conversational system that used real-time text-to-speech technologies for seamless customer interactions.
A prescription for $100 million+ forecasted savings
With more than 25,000 calls to its 1-800 customer care number each day — and with each call costing upwards of $12 — a leading health services company recognized a huge opportunity to both reduce costs and greatly improve the customer experience.
KPMG was chosen to modernize the client’s legacy interactive voice response (IVR) system, where customers followed predefined prompts along a fixed decision tree, to a conversational AI system that uses real-time natural language processing (NLP), speech-to-text and text-to-speech technologies to understand and respond to callers’ spoken requests.
Challenge: Updating a legacy system
With more than 25,000 calls to its 1-800 customer care number each day — and with each call costing upwards of $12 — a leading health services provider recognized a huge opportunity to both reduce costs and greatly improve the customer experience by modernizing its legacy interactive voice response (IVR) system.
Solution: AI-assisted customer service
We helped implement an IBM Watson conversational AI system that uses real-time natural language processing (NLP), speech-to-text and text-to-speech technologies to understand and respond to customers in a personalized, contextually-sensitive, conversational manner.
Results: An improved customer experience
Our client now has a highly-secure, easy-to-customize conversational AI system that enables its customers to engage in complex voice interactions to get the information they need without having to wait for a human agent.
Enabling technologies
- IBM Watson AI platform
- Multicloud
- Application programming interface (API) management
- Conversational AI system
- Natural language processing (NLP)
- Text-to-speech/speech-to-text
- Cybersecurity (PII and PHI protections)
- Analytics
Requirements
Massive scalability was obviously a requirement, as was HIPAA compliance.
The user experience was paramount. It had to be secure, and responses to customer questions had to be personalized and highly contextual based on customer-specific information stored in back-end systems. The extensibility of the system was also essential. If the call couldn’t be handled entirely by the conversational AI system, then a human agent would have to be prepared with enhanced tools and information. It should also be extensible beyond phone to include web and mobile applications to deliver a more consistent experience across all engagement channels.
Our response
To meet the challenge, we employed a Scaled Agile Framework (SAFe) — a system leveraging Agile, Lean and DevOps principles to implement modern software development practices at scale.
Our Scrum teams focused on eight specific areas of knowledge:
- Product management, including feature specifications
- Human-centered design, including user experience and lexical style design (voice personality)
- System architecture and design
- AI and ML, including NLP, speech-to-text and text-to-speech
- Telephony integration, architecture and design
- Software engineering and DevSecOps
- Quality assurance
- Target Operating Model definition and change management.
Ultimately, over 200 people were involved in the process, including 60 from KPMG, as well as vendor and client resources, with KPMG taking the lead.
We began by helping to evaluate nearly a dozen potential technology providers and narrowed the field to two. We helped to create a proof-of-technology for both and had each evaluated by both technology developers and nondevelopers for performance, ease of use, ease of configurability and on the quality and technical merits of each solution’s tools. We conducted a series of integration tests to develop estimates for implementing each one in the final production environment.
Once the client selected a vendor, we then implemented the solution, starting with the design of the AI agent’s persona and lexical style.
The technology
The solution was built using the IBM Watson AI platform.
This included Watson Assistant, Watson Speech-to-Text, Watson Text-to-Speech as well as IBM Voice Gateway to integrate with the IVR. Various intent recognition (Skills) models were trained to classify the intents, language-based speech model customization and customization of speech synthesis.
The solution required multiple integrations across several cloud platforms and legacy on-premise systems, as well as custom software development, including:
- Front-end systems (telephony, desktop, web and mobile)
- Channel integrations (voice agent, email, SMS and web gateways)
- Orchestration and routing (context injector, agent transfer handing, response handler, etc.)
- AI and dialog manager (intent classification, entity/keyword detection, dialog management, emotion detection, digital assistant, etc.)
- Speech-to-text and text-to-speech including emotion analysis
- API management for integrations with back-end systems and applications
- Systems of record integrations
- User authentication and store
- Monitoring and interaction analytics.
Results
With a series of two-week sprints, we were able to deliver the solution in under six months.
Our client now has a sophisticated solution that enables its customers to have extremely complex voice interactions handled by conversational AI. The system can have user-specific, context-based conversations with callers, including location-sensitive plan and benefits information—all while operating in a highly secure environment to protect customers’ sensitive personal health information. Its AI components support a wide range of caller types and demographics including age and accent differences. And after the call, a visual summary of the AI agent’s responses is delivered directly to the customer’s phone via a first-of-its-kind digital assistant.
Why KPMG?
There were several reasons the client chose KPMG.
First was our ability to handle the complete project and serve as a single point of accountability—from making the business case and platform selection through design, prototyping and implementation, including monitoring and analytics and training and marketing.
High on the list was our objectivity. Because we have the technical skills and experience to implement a wide range of applicable solutions, the client had the freedom to select the right technologies to solve the business challenge instead of molding the challenge around a specific solution.
Our ability to handle diverse technologies was also important during implementation. To answer customers’ questions, the new conversational AI system had to be connected to a host of existing information technology (IT) systems. A single call, for example, could easily move through three or four separate platforms, including Amazon Web Services, Microsoft Azure, IBM Cloud and legacy on-premise solutions.
In addition to our Ph.D.-level AI and ML talent, we also were able to bring our own proprietary AI and ML tools and accelerators, which would help to shorten the development cycles.
Finally, unless the client’s existing operating structure was reconsidered, ten senior leaders would be involved in the project’s delivery and continued operation.
Because we have both technical and business skills, we were able to help design and implement an operating model that would balance the customer experience, resourcing, oversight, funding and other requirements to help the project succeed long term.
Speed to Modern Technology
Over the last dozen-plus years, we’ve built a leading technology organization designed specifically to help information technology leaders succeed at the pace business now demands.
Unlike business-only consultancies, our more than 15,000 technology professionals have the resources, engineering experience, battle-tested tools and close alliances with leading technology providers to deliver on your vision — quickly, efficiently and reliably. And unlike technology-only firms, we have the business credentials and sector experience to help you deliver measurable business results, not just blinking lights.
Our solutions put AI to work
Reinvent your business with our GenAI solutions. Achieve greater value, drive innovation, and unlock new potential to continuously advance your future.
Healthcare is one of our focus industries
KPMG is one of the largest providers of professional services to healthcare organizations, serving more than 70 percent of healthcare organizations in the US.
We know the healthcare industry is rapidly changing, and helping our clients remain ahead of the curve with proactive thinking, planning and action is our passion.
Meet the team that helped the company improve patient experience
Our professionals immerse themselves in your organization, applying industry knowledge, powerful solutions and innovative technology to deliver sustainable results. We help turn AI into your competitive edge—increasing sales, enhancing quality, improving processes, optimizing supply chains, and more. With AI, extraordinary achievements are within reach. You can with AI.
Swami Chandrasekaran
US AI Center of Excellence Leader, KPMG LLP