Conversational AI improves customer experience
Conversational AI improves customer experience
CLIENT STORY

Conversational AI improves customer experience

KPMG helped a mortgage underwriter pilot automation for providing customers with faster access to information.

Client
U.S.-based mortgage originator and underwriter
Sector
Financial services
Project
Virtual assistant proof of concept
  • Client challenge
  • Benefits to client
  • Approach
  • KPMG insights

Client challenge

Our client is a financial services organization that works with mortgage lenders to buy single- and multi-family mortgages across the United States. It promotes mortgage market liquidity and makes funding more available to borrowers by purchasing mortgage-related securities. 

Loan seller and servicer expectations are increasing. They want faster information access. This company's leaders realized its contact center model wasn’t delivering the level of service its customers expected.

As contact center agents deal with customers each day, they constantly referenced complex, legal process documents, such as seller servicer guides, a labor-intensive and repetitive task that kept them from more strategic work. The company’s innovation lab asked KPMG to explore automation options that could improve the process. They hoped a conversational, artificial intelligence-powered virtual agent could help call center workers answer questions and guide their customers.

Benefits to client

For this client, the KPMG team developed a virtual assistant proof of concept to demonstrate the capabilities of artificial intelligence and a roadmap for training it over time. The virtual assistant proof of concept is based on typical client profiles and interaction flow for a conversational user experience. It can operate on the client’s Amazon Web Services cloud platform.

When implemented, the virtual assistant can provide 24x7x365 customer support to locations across the United States, reducing the demand for staff. It can augment human contact center workers by answering routine customer questions allowing existing staff to focus on complex questions and managing the knowledge base. Virtual agent responses will be consistent and come from a broad knowledge base.

Approach

The KPMG team worked with the client to build a conversational virtual assistant proof of concept to demonstrate the capabilities of artificial intelligence. The virtual assistant would provide the client’s customers an around-the-clock, self-serve option for answers and a better overall experience at a lower cost for the client.

To create the virtual assistant proof of concept, the KPMG team defined use cases based on typical customer use, designed software bot and client profiles, and developed design principles and interaction flow for a conversational user experience. The KPMG and client team collaborated to select the cloud platform, architect technology platform components, and construct the virtual agent through a series of Agile development sprints to operate on their Amazon Web Services cloud platform. The team also developed a roadmap outlining skills, process, and governance considerations needed to train the virtual agent.

KPMG insights

The KPMG team brought significant data analytics, technology, and industry experience to help the client determine a solution to a complex business issue. Our extensive work with data governance and cloud development experience for creating AI applications on Amazon Web Services helped fulfill the client’s needs. We drew upon technical and design know-how to successfully create and implement a conversational virtual agent the client can use company-wide, and we applied data science knowledge for constructing formative processes to maintain and train the AI system.

KPMG’s extensive work with data governance and cloud development experience for creating AI applications on Amazon Web Services helped fulfill this client’s needs.

KPMG complies with the auditor independence rules of the AICPA, SEC, PCAOB and DOL. As a result, some services described herein may not be available to our audit clients. KPMG audit clients should check with their respective lead audit partner for more information.