Accelerating Customer Experience: Principal Financial Group’s Innovative Approach to Virtual Assistants with AWS

By Mulay Ahmed and Caroline Lima-Lane, Principal Financial Group
Note: The views expressed in this post are those of the authors and do not reflect the opinions of AWS.

Accelerating Customer Experience at Principal Financial Group with Amazon Lex and CI/CD

This guest post was written by Mulay Ahmed and Caroline Lima-Lane from Principal Financial Group. The content and opinions in this post belong to the authors, and AWS is not responsible for its accuracy.

Introduction

At Principal Financial Group®, a company that handles millions of customer calls annually, enhancing the customer call experience is a top priority. To achieve this, the organization sought to modernize its customer interactions through the Principal Virtual Assistant (VA) platform, leveraging cutting-edge technologies like Genesys Cloud, Amazon Lex V2, and Amazon QuickSight.

In a previous post, we explored the overarching solution. This article zeroes in on how Principal has accelerated its VA delivery processes by implementing a sophisticated CI/CD pipeline, drastically improving the development workflow.

The Need for Modernization

Principal is a global financial company with nearly 20,000 employees dedicated to improving wealth and well-being for individuals and businesses. With 145 years of experience, the company serves roughly 70 million customers. The demand for effective self-service and routing capabilities led to the realization that key engineering opportunities needed to be addressed:

• Eliminate console-driven configurations, testing, and deployments.
• Enhance collaboration with structured version control for multiple team members.
• Accelerate development cycles through automated builds and deployments.
• Improve quality assurance via automated testing and validation processes.

Accelerating Development with Automation

By adopting automation solutions, Principal has successfully increased its development efficiency by 50% as of September 2024. This streamlined approach not only minimizes errors but also ensures that updates are consistently reliable across different environments—development, pilot, and production.

Overview of the Solution

The solution involves combining various AWS services and APIs, including:

• AWS Step Functions for orchestrating deployment workflows.
• Amazon Lex V2 for developing the voice virtual assistant.
• AWS Lambda and Amazon S3 for data processing and storage.

Code Organization and Management

Principal’s implementation utilizes Genesys Cloud as the contact center application, organized across distinct stacks:

• Bot Stack:

  • Amazon Lex V2 for bot infrastructure.
  • Lambda functions for routing logic.
  • AWS Secrets Manager for secure endpoint access.

• Testing Stack:

  • Step Functions for orchestrating testing workflows.
  • Verified test boundaries using automated testing gates.

• Analytics Stack:

  • Amazon Athena for querying data.
  • Amazon QuickSight for powerful data visualization.

CI/CD Workflow with Amazon Lex

GitHub acts as the source control repository, streamlining the CI/CD pipeline. Here’s how the process works:

1. A developer clones the repository and branches off for changes.
2. Changes are made, and a pull request is created.
3. The developer either merges with the main branch or deploys locally.
4. The automated pipeline runs various tests: linting, unit tests, and integration tests.
5. Upon successful completion, the code is prepped for production.

Automating the Test Workbench

Instead of relying on manual testing, Principal integrated automated testing through AWS services. The pipeline enables automatic uploads of test datasets, triggering comprehensive testing through a Step Functions state machine.

Key Practices for Test Set Management

To maintain consistency and effectiveness in testing:

• Version-control test set files linked to their respective bot versions.
• Create and regularly update golden test sets, incorporating real customer utterances to boost intent recognition rates.
• Deploy versioned test data with each bot deployment in non-production environments.

Conclusion

In summary, Principal Financial Group has harnessed the power of automation and AWS—particularly Amazon Lex V2 and CI/CD practices—to refine its development and testing processes. This transformation has led to significant reductions in testing time and increased deployment reliability, ultimately enhancing customer service capabilities.

By embracing these capabilities, Principal sets itself up for ongoing enhancements and a robust infrastructure for the Virtual Assistant ecosystem. For a deeper dive into these services, check out "Evaluating Lex V2 Bot Performance with the Test Workbench."

About the Authors

Mulay Ahmed is a Solutions Architect at Principal, specializing in architecting complex enterprise-grade solutions, including AWS Cloud implementations.

Caroline Lima-Lane is a Software Engineer at Principal with extensive experience in the AWS Cloud space.

This post illustrates how innovative technologies and effective automation strategies can fundamentally improve customer interaction processes in the finance industry, setting a new standard for service excellence.