Capital Group
Retirement Planning & AI/ML

Industry:

Finance

Role:

Sr. UX Designer, IC

Challenge:

Update and modernize internal facing account management tools.

Users:

  • Call Center Reps
  • Account Managers
  • Flex/Temp workers
  • Non Technical

Methods & Technologies:

  • Contextual Inquires
  • Prototyping
  • User Validation via Pilot Testing

Capital Group is an American financial services company that ranks among the world's oldest and largest investment management organizations, with over $2.6 trillion in assets under management.

I joined Capital Group as a member of the UX team that was tasked with improving the user experience of internal applications for account representatives. Our focus was part of an ongoing effort to migrate from their legacy Application of Record to a new platform.

whiteboard exercise

FI Processing Tools User Pain Points

One benefit of designing for internal users is the ability to gain relevant, actionable feedback quickly. I spent time sitting with users as they took calls, meeting with users to find what pain points they were experiencing with their current application stack, and how changes were received in early adopter test environments.

Working with Project Sponsors and Stakeholders we would often audit and evaluate existing tools to determine opportunities for improvements, and the teams would prioritize accordingly.

AI/ML Explorations

Common errors during digitization coupled with security and compliance meant that highly trained account reps spent additional hours reviewing documents for accuracy before processing them. Flex staff could not review the documents in entirety due to sensitive client account information.

Using time as a key metric we explored sending singular fields that failed during the digitization process to multiple associates, including Flex staff, instead of sending the entire form for review to a single associate. Using this data to train AI/ML to correct common digitization errors during the digitization process would also decrease the rate of human error and expedite the time to process accounts.

AI/ML input for common typos
Common typos can be corrected out of context by entry level staff, preserving customer account privacy while freeing up account reps for processing.

Common typos can that appear obvious to people can be problematic for system validation and processing. In this example, "Palace" is spelled correctly, but a person familiar with cities in California would recognize quickly that the customer intent was actually "Palos Verde". Taking these small corrections into account, the ML engine over time will suggest proper corrections.

Retirement Plans

As part of the ongoing effort to migrate from their legacy Application of Record to a new platform, Retirement Plans was a feature that caused a lot if friction and was a time sucker. The AOR is over 20 years old, lacking in functionality, as well as the ability to update the code base. The legacy system was poorly built and lacking in best practices for UI/UX, but the users were entrenched in the mental models and muscle memory of the outdated system that facilitated their daily work flow.

Retirement plans were mailed into a processing center, digitized, and then sent to an account rep to manually enter into the retirement planning tool. The tool I had to design needed to utilize the Appian framework.

Visual document mark-up tools and notes functionality allowed reps to quickly interact and resolve questions in a timely manner.