Case Study
Mobile App · Internal Tools · AI Integration
Role
Lead designer
Redesigned a safety-critical reporting platform across legacy constraints, mobile workflows, and later AI-assisted workflows.
Over 6 years, I owned this product from initial design through AI integration.
+102% report submissions in the first week
Framework adopted across multiple frontline roles
First United app to integrate AI voice technology
Framework now informing a company-wide safety reporting initiative
THE CONTEXT
Safety reporting at an airline is a core operational system used by 90,000 employees to protect passengers and each other across every flight.
But the existing system hadn’t evolved with frontline work. Built on legacy infrastructure and designed for desktop use, it created friction for mobile-first, time-constrained employees, over 70% of whom work on the move using mobile devices, and contributed to a compliance gap in the 48-hour reporting requirement.
Before designing anything, I led requirements gathering and task analysis with airport operations employees to understand how reporting worked in practice versus how it was assumed to work.
UNDERSTANDING FRONTLINE WORKFLOWS
Initial concepts relied on employees selecting flights from their schedules. However, research revealed that gate agents and ramp technicians are frequently reassigned to flights at the last minute, making scheduled workflows unreliable in real operational contexts. I had to design a way for employees to manually add flights.
Usability testing showed that standard progress trackers created friction in dense, multi-step mobile workflows by taking up too much screen space. I introduced a lightweight transition pattern that preserved orientation without reducing usable space. The pattern was later adopted into the Employee UX mobile design system as a reusable component for future form experiences.
Slide from usability testing readout with stakeholders.
THE PIVOT
Midway through the project, after two rounds of usability testing, the product shifted to launch first for flight attendants.
We had already begun restructuring the single-page desktop form into a modular mobile experience, but the information architecture was still evolving as early workflows were validated.
Rather than restart, I adapted the emerging system into two distinct workforce models.
Airport operations roles (gate agents, ramp technicians) operate on unstable schedules, making schedule-based flows unreliable. Flight attendants operate on structured schedules, making them effective.
I formalized a reporting taxonomy across flight, station, incident, conditions, and narrative domains to support both contexts within legacy system constraints.
The resulting framework allowed the reporting experience to adapt interaction patterns to the needs of each employee group.
FLIGHT ATTENDANT EXPERIENCE
Dictation
Native voice-to-text to support
on-the-go reporting in time-constrained environments
Progressive disclosure
Conditional questions that surfaced only when relevant to reduce cognitive load in long-form reporting
Review step
An editable summary before submission to allow users to refine and validate their report
Some features were deprioritized for the initial release, but user feedback later supported their re-prioritization in subsequent iterations.
5 years later…
The reporting framework remained in active use across United’s workforce and later evolved as part of a broader initiative to introduce AI-assisted employee workflows. MSA became the first reporting experience to integrate with aiola, a voice AI platform that United partnered with to convert dictated narratives into structured report fields.
RESPONSIBLE AI INTEGRATION
Designing AI transparency without disrupting workflow
United’s Responsible AI team required two conditions before approval: a persistent transparency disclosure and a user feedback mechanism. These requirements conflicted with product goals around minimizing friction in the reporting flow.
I placed the AI disclosure directly beneath the "tap to speak" interaction, preserving the visibility of the primary action while maintaining access to transparency details throughout the reporting flow. Inline messaging also clarified when report content had been generated using AI.
FEEDBACK MECHANISM
A conditional pattern that aligned governance requirements with the natural end of the reporting experience
If AI dictation was used, the review screen surfaced a clear indicator that the narrative was AI assisted.
The feedback mechanism was embedded directly into this step, aligning governance requirements with the natural end of the workflow.
This approach balanced Responsible AI requirements with product constraints while preserving a low-friction reporting experience.
IMPACT
The framework originally built for MSA now informs United's company-wide safety reporting initiative
The framework I designed in 2020 is now being carried forward into a company-wide initiative to rebuild safety reporting infrastructure across United's entire workforce.
OUTCOMES
102% increase in report submissions in the first week
Reporting framework scaled across multiple frontline employee groups, including gate agents, ramp technicians, pilots, catering, and cargo teams
Employee-submitted reports directly informed operational improvements, including infrastructure updates at DIA gates to improve passenger comfort