UX Research
Design System
User Interviews
Prototyping
Airudi
🔒
Details have been generalized and visuals removed to protect confidentiality.
Objective
0->1 Design for AI Nurse Assignments
Timline
5 months (July 2025 – Dec 2025)
Team
1 Product Owner, 2 Designers, 2 ML, 4 Engineers
Contribution
Competitive research, interviews, surveys, JTBD, field observation, ecosystem mapping, as-is/to-be workflows, high-fidelity AI prototyping, usability testing, dev handoff
Problem
Frustration in percieved inequitable workload distribution
Nursing staff described burnout as being exacerbated by heavy administrative workload and a perception that workload distribution was not always equitable. Managers, meanwhile, had limited visibility and tools to coordinate assignments efficiently, making it challenging to balance workloads fairly.
💭
How might we reduce administrative burden and improve trust in workload distribution without increasing complexity or cognitive load?
Objectives
⚖️
Support balanced workloads and improve perceived fairness/trust
👣
Reduce manual coordination steps and tool switching in assignment workflows
⚡
Enable faster adjustments during shift changes and exceptions
🧑🏽💻
Ensure the experience remained usable under real-world constraints (interruptions, time pressure, role-based coordination)
Constraints & Considerations
We encountered several constraints during the project
We therefore prioritized high-fidelity prototyping early to validate workflows quickly and reduce implementation risk.
⌚
Aggressive timeline
with limited room for scope creep
🔌
Integration dependencies
outside our direct control, impacting data availability and update cadence
⚙️
Operational realities
(interruptions, shift handoffs, rapidly changing conditions) requiring speed and resilience
🤝
Trust requirements
for AI-assisted workflows: adoption depended on transparency and user control
Process
The process at a Glance
Competitive Landscape Synthesis
Competitive landscape review to identify opportunities and pitfalls
I conducted a competitive review across a broad set of relevant products and adjacent solutions.
🎯
We identified industry patterns, trends, and differentiation opportunities - focusing on workflow support, trust, and operational fit - to guide early design principles and anticipate adoption pitfalls.
Research
Interviews, Surveys & Field Observation
🔍
Methods
5 in-depth interviews
14 survey responses, primarily from frontline staff
Weekly touchpoints with nurse managers to validate feasibility and operational constraints
Field observation to understand the lived reality of coordination and assignment work
NOTE: Quotes are paraphrased from multiple conversations. Details have been generalized to protect confidentiality.
It takes a significant amount of time to gather the information needed to create fair assignments.
🔍
Key insight: Manual and fragmented workflows
Across interviews and observation, we saw that key workload signals were often calculated manually using disconnected tools. This increased coordination time, introduced inconsistency, and created information gaps.
north star
JTBD Framework
We translated research insights into a JTBD framework to keep the team focused on core outcomes and avoid over-optimizing for smaller details.
As-is & To-be
Ecosystem Mapping → Future Blueprint
As-is workflow mapping
Each step and resources used
System touchpoints
Manual workarounds
Sources of friction and repeated effort
To-be service blueprint
User actions
Supporting system behaviors
Constraints & dependencies
Expected simplification of coordination and reduction in unnecessary effort compared to the as-is flow
✅
Aligning on the blueprint upfront ensured the team shared the same operational goals before building
It reduced time, steps, & cognitive load, as well as preserving safety and control.
Decisions
Key Design Decisions
01
Consolidate manual calculations into the platform
Manual calculations created friction, so we explored ways to centralize effort and reduce duplicate work across the workflow.
02
Reduce information asymmetry to improve trust
To address trust challenges caused by uneven access to information, we focused on making workload-related context more visible and understandable across roles.
03
Design for exceptions and rapid change
Because conditions change during a shift, we designed for fast edits and clear paths to adjust assignments when needed.
04
Align automation behaviors with clinical expectations
I partnered closely with engineering and applied AI teams to translate research insights into system principles and edge-case considerations.
Prototyping & testing
AI Prototyping & Usability Testing
I facilitated weekly alignment sessions to validate scope, interactions, and edge cases. Prototypes were shared with additional target users to gather formative feedback, which informed iterative refinements prior to development handoff.
Purpose
Validate an early concept and identify usability risks prior to implementation.
Iterations made
Improved visibility of key actions
Adjusted UI formatting to improve scanability
Clarifying how exceptions and adjustments could be handled
🔒
Visuals have been removed to protect confidentiality.
outcome
In testing and iterative feedback sessions, users responded positively
Better workload transparency
Transparency cues reduced uncertainty and improved perceived fairness
Fewer manual steps
Reduced tool switching and eliminated reliance on memory for contextual cues
Better exception readiness
Stronger support for mid-shift changes through fast edits and override paths
I facilitated weekly alignment sessions to validate scope, interactions, and edge cases. Prototypes were shared with additional target users to gather formative feedback, which informed iterative refinements prior to development handoff.
reflection
Closing Thoughts
With more time, I would have broadened usability testing across additional roles and contexts, explored edge cases over longer work periods, and examined more configurability to better support variation in workflows and environments. Early formative feedback during design reviews was directionally positive, but outcomes were not measured quantitatively.
Prioritizing Scalability
I realized that scalability needed to be treated as a core requirement rather than a future consideration. This shaped how I approached reusable patterns, system alignment, and designing for adaptability beyond a single initiative.
