Executive Summary
The goal of this project is to leverage cutting-edge AI technology to deliver a best-in-class customer service experience across all retail products in a consistent manner, predictable, and measurable manner. Current state involves disjointed GenAI solutions across retail product lines and inconsistent knowledge stores. The initiative focused on integrating AI-driven solutions to enhance efficiency and personalize customer interactions. Using GenAI capabilities, HR Block aimed to provide seamless support across voice, chat, social, SMS to ensure consistent and high-quality service experiences for customers.
I represent the lead technology partner for this effort, directing a large cross functional team that includes internal teams representing GenAI, Enterprise Architecture, Client Services, Software Development, and Sourcing. I lead all aspects of this project includes business case defense to C-Level, RFP for Vendor Selection, Implementation of the solution, and on-going in-house support of the solution.
This effort includes strategically building up new internal core-competencies, such as Knowlege Management (used to train AI Models), customer sentiment analysis, bot flows being used across multiple communication mediums (voice, chat, sms, social, ...) all to be used seamlessly alongside existing support technologies.
Strategic Aspects
- Development of a comprehensive technology roadmap including alignment of existing contract (removing overlapped services and establishing new services) and budget forecasting (+$4M).
- Establish an internal knowledge management practice across retail products and operational processes in order to support GenAI services. Key aspect was internal control of this content to support the ability to pivot GenAI partners as this technology is expected to change rapidly.
- Focus on delivering a consistent and seamless customer experience across all retail products using GenAI.
- Partnering with Executive Level Leadership over Customer Support areas to achieve strategic KPI's including (1) Increase in Containment rate, (2) Reduction of mean-handle-time, and (3) increase rate of single contact resolution, and (4) CSAT improvement.
Tactical Aspects
- Integration of AI-driven solutions for personalized customer interactions across all support channels (Phone / IVR, Chat, Social, SMS, etc)
- Development of a unified communication framework across all product lines for customer support agents to support retail customers.
- Establishment of a knowledge management team to train AI Models. Mapping of HR Block products and business process.
- Training and skilling of human agents to work collaboratively with AI systems. Aspects included "suggested replies to customer in real time", "Next Best Action", and integrated via RESTful API to HR Block internal platforms.
- Implementation of scripted responses and automated processes to streamline customer support.
The goal of this project is aimed to transform its customer service landscape for our tax customers by harnessing cutting-edge AI technology and implementing robust knowledge management strategies the result of which will provide fast, accurate resolutions to customer service issues across commonly used communication channels while reducing internal support staff.
Akey aspect of this project was the internal effort to establish internal documentation to train AI models. This Knowledge Management aspect was housed internally and allowed HR Block to create a foundational set of data to train models and by retaining internal control (not houses by the service provider) allowed long term flexibility as GenAI technology improves over time.
Initial pilot of this project included GenAI response over a single product line integrated into our voice support channel. Issues around latency due to multiple integrations (Speech to Text -> GenAI Response -> Text to Speech) and customer meta data cross various disjointed services created challenges and highlighted the need for a single cohesive solution and a commitment accurate, detailed, well tagged content to train the model was critical. This project represented learning from this pilot.
