UI/UX Design Flow:
Business+User Goal ---> Ideation --->UX Research &Analysis ---> Wire-framing---> Usability Testing ---> Client Demo
​
​
​
UI/UX Design Flow:
Business+User Goal ---> Ideation --->UX Research &Analysis ---> Wire-framing---> Usability Testing ---> Client Demo
​
​
​
Ascension
OVERVIEW
Lead designer for patient chat and support at Ascension, driving AI-powered, research-led solutions that improve care navigation and scale across the enterprise using Salesforce.
Role: Lead Product Designer & Researcher – Patient Chat & Contact Center (Ascension)
Focus: AI integration, workflow optimization, and cross-platform support
Stakeholders: PMs, Eng Leads, AI/ML, Clinical Ops
Users: Patients, Caregivers, Care Navigators (referred to as CNs, support agents), CX Specialists
Duration: 1 yr 8 mo (ongoing)
Tools: Figma, FigJam, dscout, UserZoom, Amplitude, JIRA. and Salesforce Health Cloud
Goal: Leverage a phased approach to integrate AI into Care Navigator workflows and the patient experience streamlining their workflows, reducing manual effort, and enabling faster, more personalized care.
T H E A P P R O A C H
Step 1: initial Discovery and Research
Defining success
​
In collaboration with Product leadership, and guided by data analysis and the broader product vision, we defined the following goals:
-
Drive a 10% increase in chat adoption by shifting patient support interactions from phone to scalable, digital-first channels
-
Reduce average call wait times by at least 50 seconds to improve care access and reduce operational load on Care Navigators
-
Design a unified, intelligent support experience that enhances both patient and CN workflows by streamlining workflows, reducing manual tasks, and surfacing timely, actionable context.
​
Understanding Challenge
To improve the Care Navigator (CN) experience, I needed a deep understanding of their daily workflows, patient interactions, tools, and pain points. I conducted user interviews followed by contextual inquiry sessions to uncover insights and build a clear picture of their end-to-end experience.
Gathering data
​
I used Amplitude and Salesforce tracking to analyze Care Navigator interactions focusing on average response times and the types of patient queries received. This helped uncover common trends, patterns, high-frequency topics, and areas where support tools could be optimized to improve efficiency and consistency.
Please note: This data is confidential and won’t be shared to respect privacy and maintain trust.
User interview goals
-
To gain a better understanding of our care navigators' needs and expectations.
-
To identify any challenges they may be facing.
-
To explore potential opportunities for improvement and innovation.
User interview questions
User interview outcomes
User interviews helped in understanding Care Navigator typical tasks, schedule, and general painpoints.
​
-
Care Navigators work in shifts and manage incoming chats and phone calls from patients. They assist with centralized scheduling, provider lookup, technical issues, and general support questions.
-
They often handle multiple chats simultaneously and are expected to respond within strict time limits.
-
Their tools include Salesforce MIAW chat, Athena, Dash, and ServiceNow, typically used across multiple open screens.
-
The role can be high-pressure and emotionally taxing, especially when dealing with frustrated or angry patients.
Typical care navigator setup
CN sharing struggles with quick, accurate responses.
"The one thing I would absolutely love is: before we used to have Grammarly and it would fix your mistakes for you.
I would love love love for that to be an AI put into Salesforce.
It would make editing so fast because it would fix it for you and then you wouldn't have to sit there and go back and fix your mistakes."
Grammarly was removed due to security concerns, leaving a gap in the Care Navigator (CN) workflow by eliminating a seemingly small but essential tool that helped save time and ensure response quality without replacing it.
Care navigator (CN) shadowing
​
I conducted contextual inquiry with 6 care navigator, 2 in person at the contact center and 4 remotely. Primary goal was to understand pain points, any workdlow and experience of care navigators.
​
Key Observations
​
-
Multi-Screen Workflow
Care Navigators work across multiple screens simultaneously:-
Screen 1: Salesforce MIAW chat
-
Screen 2: EHR systems like Athena, Dash, Epic
-
Screen 3: Knowledge base, team messenger, and reference materials
This constant context-switching leads to cognitive overload and slower response times.
-
-
Pain point - High-Pressure Response Time
Navigators have just a few minutes to resolve each patient chat or call. Every interaction is expected to be a one-touch resolution, adding time pressure and urgency. -
Pain point - Inefficient Messaging Tools
To respond quickly, they rely on:-
Quick text templates
-
Self-created spreadsheets of common replies
-
Manual typing of responses
They frequently expressed concerns about grammar accuracy and professionalism, which adds stress and slows down interactions.
-
Understanding Salesforce capabilities and licenses
​
Understanding Salesforce capabilities and licensing was essential to ensure proposed solutions aligned with platform limitations, supported scalability, and could be implemented without friction.
Strategy with Product, Operations, and Salesforce
Step 2: Ideate and Solution
Ranking and prioritization exercise with Care Navigators and stakeholders
​
Leveraging Salesforce AI enabled smarter, more efficient support for Care Navigators by streamlining workflows, suggesting context-aware responses, and reducing manual effort. This helped improve response accuracy, decrease handling time, and ease the cognitive load during high-volume interactions. But we wanted to know what features and what to prioritize keeping Care Navigators in the center.
Ranking exercise with CNs
We worked with Care Navigators to understand which features resonated most with them and what would be most helpful in their day-to-day workflows. We met with 5 CNs with varying seniority levels for the ranking exercise.
Outcomes
We worked with Care Navigators to understand which features resonated most with them and what would be most helpful in their day-to-day workflows. We met with 6 CNs with varying seniority levels for the ranking exercise.
Heatmap showing the CN ranking of the AI features based on impact
Impact feasibility matrix
Prioritization exercise with product and stakeholders
​
We ran an impact-feasibility matrix with product and stakeholders, using Care Navigator insights to prioritize features. Reply recommendations stood out, but due to Salesforce licensing limits, case classification was initially ranked higher. After further discussion, we aligned on implementing reply recommendations.
Hi-fi Wireframes
When I started designing the mocks I based them on the findings of the contextual inquiry, user interviews, and Salesforce Lightening Design System. All these steps help us understand and create designs.
Salesforce AI recommends replies based on the ongoing conversation with patients
Care navigators can directly post replies or edit them
Step 3: Prototype and testing
Prototype/POC
​
I collaborated with Engineering to build a proof of concept that allowed us to test the feature in a live, HIPAA-compliant production environment. We released it to a small group of Care Navigators and gradually opened access to patients, enabling real-time user testing and iterative improvements to the AI-driven experience.
UXQA and User Testing
​
UXQA
As part of launching the AI-powered feature, I implemented a focused UXQA process to ensure design integrity. Collaborating closely with Engineering, I reviewed the live proof of concept against design specs validating interaction patterns, visual consistency, accessibility, and overall flow. This helped us catch and resolve issues early, ensuring Care Navigators had a smooth, intuitive experience during real-time testing.​​
User testing
The purpose of this research is to understand how Care Navigators currently use the Salesforce (SF) experience while chatting with users and to evaluate the effectiveness and usability of the new AI-generated reply recommendation feature.
"Sometimes I get a little nervous with the grammar. I like how [feature] had grammar and even if I had to adjust anything, it was very simple to do. So, grammar-wise, it made me feel comfortable."
Care Navigators praised the recommended replies feature for resolving the communication pain point identified earlier.
Key Observations
​
-
Highly Recommended:
CNs gave an average score of 8.4/10 for recommending the feature to your colleagues, showing strong confidence in its value. -
Quality & Relevance:
-
Recommendation quality rated 3.8/5 — generally "Good," with some room to improve consistency.
-
AI replies alignment scored 4.0/5 — mostly matched what CNs wanted to say, with consistent positive feedback.
-
-
Workflow Impact:
-
Efficiency boost with a 4.2/5 rating — made work faster for most, none reported it slowed them down.
-
User Comfort:
-
4.8/5 comfort rating — CNs felt very comfortable and confident using the feature, with minimal variance.
-
Summary: Strong reception, meaningful workflow benefits, and high ease of use — with opportunities to refine recommendation consistency.
It’s not done yet.
​
I continue to partner with product, engineering, and stakeholders to refine the feature based on Care Navigator feedback, with plans to expand the rollout across a broader group.
​
Key areas of focus:
-
Tone & Empathy in AI Responses:
Current training data lacks warmth and conversational tone.
Recommendation: Align training data with Ascension’s brand voice to support trust and empathy in patient communication. -
Training Data Gaps (Quick Text & Knowledge Base):
Quick Text and KB content aren’t reflected in model recommendations.
Recommendation: Integrate both to improve response accuracy and relevance. -
Thumbs-Down Feedback UI:
The current experience isn’t intuitive for CNs.
Recommendation: Redesign feedback flow to improve clarity and engagement.
​
This project remains a living, collaborative effort and I’m excited to continue shaping it into something that truly supports both CNs and patients.
Project Management
-
Regular cadence with product leadership, operation leadership and internal teams
-
Alignment on upcoming design deliverables
-
Alignment with internal stakeholders, like product managers, and developers
-
Maintaining a design backlog
-
Review meetings cadence
Reflection
This is still very much a living project one that continues to evolve as we explore the role of AI in supporting patient care. From the start, we partnered closely with Ascension’s ethics and governance teams to ensure every decision met rigorous standards for privacy, security, and compassion. Working with AI in healthcare requires more than innovation; it requires responsibility and we’ve approached every step with care.
​
One of the most meaningful contributions I brought to the team was introducing a data-driven, design thinking process that hadn’t existed before. We began using Amplitude to track behavior and measure success, which allowed us to tie feature usage directly to Care Navigator feedback and patient outcomes. This shifted the conversation from assumptions to insights and empowered us to iterate with purpose.
​
Bringing AI into our workflow is exciting but we’re moving intentionally. By grounding our decisions in real-world data and continuing to test in collaboration with CNs, we’re building something more than a feature. We’re creating a tool that supports real people, in real moments, when care matters most.