To meet our tight deadline, I knew the first step was understanding what our system could handle. I immediately collaborated with the engineering team to evaluate the technical feasibility, identify system limitations, and ensure compatibility with proposed solutions. This partnership allowed me to focus on researching options aligned with our scope and timeline. By gaining a thorough understanding of our technical debt and constraints, I was able to streamline my design process, discarding impractical ideas early on and transforming ambitious visions into concrete, workable goals that could be implemented effectively.
We also uncovered a HIPPA concern regarding saving user answers. As an HR platform, we can save and store information such as first and last name, SSN, date of birth, and addresses for each family member for which the employee designates benefits. Still, for the healthcare support AI tool to work effectively, it needs to know health-related questions. How do we hold this information long enough to populate a result and allow the user to select, save, and wipe it afterward?
Handling storage information was a sensitive topic, but I recognized that a smooth user experience required at least retaining user data within the session. This approach allowed users to navigate back and forth during enrollment without re-entering their responses. After further investigation, we confirmed that saving previous recommendations and aggregated statistics would not violate any regulations, provided they didn't include any directly identifiable personal information.
During this time, I also documented specific scenarios where users would interact with the open enrollment platform and need benefit recommendations. Each scenario carried its own workflow, which helped us notice any possible similarities we could combine. The main benefits were medical, life, voluntary additions, and retirement investments. Other benefits, such as dental, vision, etc., need statistical recommendations, not personal ones.
Key Takeaways:
- We needed to receive and analyze data on past medical claims, lifestyle inputs, and budgets to recommend the most suitable health insurance plan.
- Our system was able to integrate the AI API into the existing system.
- We needed the platform to perform and display tasks like personalized recommendations or plan comparisons.
- Ensuring secure data transmission (e.g., encryption) and compliance with regulations like HIPAA was critical.
- A complete workflow change was not doable and would require it to be an independent project.
- A completely new design look was not doable within this project. Still, we could incorporate updated UI elements without affecting the legacy design.