About 1 in 7 people in the United States experience some degree of hearing loss - T-Mobile's frontline account of turning an "edge case" into a product everyone benefits from
The edge case is everyone ft. Willie King & Lex Liley | Config 2026 · Willie King
20 min total·Actually worth watching closely: ~4 min·2 must-watch clips
- 0:00 – 4:10Listen
The proposition: the edge case is everyone
The opening sets out the core proposition: not being able to communicate in the moment is the daily reality for over 11 million deaf and hard of hearing Americans, and the phone is a barrier by design - this is nowhere near a niche problem.
The so-called edge case is far larger than it looks, and designing for accessibility ripples out to make things better for everyone
This stretch is spoken setup of the problem with no key visuals - fine to listen to while doing something else▶ Jump to 0:00Speaker · Willie King - 4:10 – 7:30Listen
Pain on both ends: the call is stuck at both sides
Willie's story of why the phone as it operates today doesn't support the calls that matter - communication barriers aren't theoretical in a house where he and his wife are raising two deaf daughters - and why T-Mobile decided to build toward a world where communication actually works, rather than one people have to adapt to.
Everything around us, especially communication, assumes you can hear and speak; the barrier sits at both ends of the call, so the design has to answer both
Still mostly argument, with no visual dependency marked in this chapter - catching the problem definition is enough▶ Jump to 4:10Speaker · Lex Liley - 7:30 – 10:28Listen
Learning one: design for a range of preferences, not a persona
Communication across this community is incredibly diverse - speaking in your own voice, typing only, listening with transcription as support - and preferences can shift partway through a call. People with a cochlear implant who don't sign are part of who you're designing for too.
Being deaf or hard of hearing does not mean silence; the target is a shifting spectrum of preferences, not one typical user
Mostly argument in this stretch - the interface footage worth watching doesn't start until 628s▶ Jump to 7:30Speaker · Lex Liley - 10:28 – 12:35Watch
In practice: three interactions in one interface
A walkthrough of the call interface that user testing converged on: type your response and have it spoken to the other caller, speak your response and read what comes back in real time, or hear the response and reply through your keyboard - three patterns in one interface, switchable without leaving the conversation.
The balance between flexibility and clarity comes from keeping settings changes inside the conversation instead of pushing people into a settings screen
Both LLM-flagged visual moments sit in this chapter, with the real interface on screen - audio alone misses the product entirely▶ Jump to 10:28Speaker · Lex Liley - 12:35 – 14:33Watch
Learning two: accessibility is a practice - the dark mode lesson
Dark mode as the clear example that accessibility can't be delivered once: the first instinct of simply inverting the colors produced the wrong result, and the fix was defining text and background pairings that hold up across both light and dark modes, then having the accessibility team pressure-test every scenario.
Even a design system built with accessibility in mind will have real users surfacing needs you hadn't anticipated - it's an ongoing practice, not a feature toggle
The side-by-side color comparison appears at 754.92s, and the visual evidence is far more direct than the spoken description▶ Jump to 12:35Speaker · Willie King - 14:33 – 17:30Listen
Learning three: walk with your users, don't just design for them
Willie, a deaf team member and the visionary behind the project, was in the room throughout, which changed how the team operated: cameras always on, introducing yourselves every time you spoke, side chat cut out. Plus the immersive simulation - designers in separate rooms, one speaking while the other typed silently, living the call themselves.
Insight and empathy are not the same thing; a practitioner who doesn't live it has to change how they work - the method matters less than the mindset
No visual dependency marked in this chapter; the value is in the spoken detail about how the team collaborated, so it rewards focused listening▶ Jump to 14:33Speaker · Willie King - 17:30 – 20:08Listen
Quick replies, and onboarding as an emotional experience
After prototype testing exposed the pain at both ends, one-tap quick replies kept both parties in the conversation. Onboarding was then designed as an emotional experience - choose the gender of your synthetic voice or go completely neutral, walk the core features in a low-stakes way - so the call feels like yours before you've made one.
The best features tend to come from feeling a user's real frustration rather than from a requirements list; onboarding should build a sense of ownership before it chases efficiency
The closing is narrative and takeaway-driven with no flagged visual moments - the conclusions carry it▶ Jump to 17:30Speaker · Lex Liley