Sponsored by Hiya — Global Leader in Voice Security and Caller Trust
Overview
In 2030, AI will be embedded in how patients reach care, yet the phone will remain the clinic’s front door. The question is no longer what AI can do, but how it should behave to be trusted in sensitive contexts. Partnering with Hiya, I explored outpatient clinics as a domain to apply its strengths in voice and identity.
Scope
This project had two phases. During the capstone, our team explored how AI could rebuild patient–clinic trust through research and iterative concepts. Afterward, I synthesized the learnings into a set of trust guardrails and a Framework of Trust, creating a transferable model for AI communication beyond healthcare.
Outcome
Designed an AI receptionist and a staff assistant that triage requests, reassure patients, and pass context to maintain clarity, continuity, and trust in patient–staff communication.
My Role
Defined the design vision and alignment strategy to establish how trust should shape human–AI communication, and developed a Framework of Trust to guide AI design across domains.
Team
2 Designers, 1 Researcher
Timeline
Jan – Aug 2025
Platform
Healthcare Communication (Voice AI)
The Solution at a Glance
Case Study Map
Where it Began
Exploring Healthcare as Hiya’s 2030 Entry Point
Reframing where trust matters most
Hiya’s reputation in caller trust and voice security protects millions each year. Building on that base, we reframed the question from what to build next to where trust matters most for voice AI by 2030.
200M+
Active Users
150B+
Calls Analyzed Annually
#1
in Caller Trust
Clinic calls as the first test of care
Outpatient clinics remain phone-first, with 30% of calls for scheduling and no-shows. As care’s first and most emotional touchpoint, they reveal where empathy and accountability slip. With Hiya, we explored how voice AI could rebuild confidence in these moments.
My Role
Framing How AI Builds Trust in 2030 Healthcare
During project: Driving vision and alignment
Led the design vision and alignment for Hiya’s voice-AI exploration in 2030 healthcare, framing future scenarios and prototyping patient–AI–staff handoffs to define how trust should shape human-AI interaction.
After project: Systematizing trust principles
Extended the project beyond healthcare to transform fragmented insights into a cohesive Framework of Trust that defined principles for clear and confident human-AI communication.
Research
In 2025, patients waited on hold. By 2030, AI takes routine, providers bring reassurance
My Impact
Led tension synthesis to reveal how trust broke across patients and staff.
Framed and ran a workshop that turned tensions into shared focus.
Approach
Uncovering Where Trust Breaks and Why It Matters
We noticed trust often broke not at errors but at moments of silence between patients and staff. Through interviews and workshops, we learned confidence grows when clarity and reassurance move together.
Current Breakdown
2025: Broken Calls, Frayed Connection
We realized breakdowns came not from mistakes but from moments when patients and staff lost each other’s attention. Missed calls and repeated questions showed how small gaps quietly wear down care.
Reach Clinic: Patients on hold, staff overloaded
Patients wait on hold while staff juggle multiple ringing lines and front-desk tasks. Both sides grow anxious as patients feel ignored and staff feel stretched thin.
Get in touch: Patients repeat, staff lack context
Patients repeat the same details across calls while staff dig through disconnected records. Context is lost, conversations stall, and neither side feels in sync.
Unanswered concerns: Patients dismissed, staff overloaded
When messages go unanswered, patients lose clarity and confidence while staff under pressure give rushed or vague replies. Speed replaces care and trust slips away.
Key Tensions
Mapping the Boundaries of Human–AI Collaboration
In a workshop with Hiya’s teams, we reframed patient–staff friction into shared human–AI tensions. Everyday misalignments began to trace the deeper boundaries shaping emotion, understanding, and control.
Clarity vs Ambiguity: Confusion breaks credibility
Everyone needed to know who they were talking to and what came next. Patients wanted clear handoffs, staff needed visibility into AI’s role, and Hiya sought clarity without overexplaining.
Reliability vs Fragmentation: Broken systems break confidence
Everyone needed communication to stay consistent and accurate. Patients wanted info to carry over, staff relied on context, and Hiya worked to route the right details to the right person.
Efficiency vs Empathy: Speed challenges sincerity
Everyone wanted faster, more caring communication. Patients expected quick reassurance, staff needed efficient tools that still felt human, and Hiya aimed to prove automation could feel warm.
Oversight vs Autonomy: Control adds cognitive load
Everyone wanted AI that helped without overstepping. Patients needed to feel humans stayed in charge, staff sought oversight without extra work, and Hiya built accountability that stayed light.
Market Scan
Everyone Solved for Care, Just Not Together
By 2025, we saw each product fix a part of communication: clarity, empathy, context, or oversight, but rarely all at once. Care advanced in fragments, never as one connected system.
Future Roles
2030: AI Handles Routine, Humans Carry Empathy
Building on 2025 tools that automated fragments of care, we explored how responsibility might rebalance by 2030. Routine shifted to AI, while reassurance and judgment remained human.
Opportunity
Designing Seamless Human–AI Handoffs
As routine calls move to AI, staff face fewer but weightier moments of judgment. We saw the next opportunity in designing handoffs where confidence flows seamlessly across patients, AI, and staff.
Frame the Problem
Design Probes
Probing Communication Across Boundaries
To move from foresight to evidence, we turned each workshop tension into a design probe. Each tested a key handoff to see how confidence could travel through changing roles and context.
Evolving the Design
Examining how AI and staff share calls while keeping care steady and reassuring
My Impact
Shaped patient-AI dialogues that turned empathy into clearer first contact.
Defined staff handoff and escalation to balance empathy with oversight.
Phase 1
Exploring Clarity and Reliability in Handoffs
Breakdowns often surfaced at the handoff, when patients moved between AI and staff. We examined how clearer context and steadier routines might keep communication consistent across shifting roles.
Testing how clarity and reliability hold across roles
We translated workshop tensions into two probes: Contextual Handoff for clarity, passing emotional context between callers and staff, and Routine Automation for reliability, maintaining flow through follow-ups.
Designing clear and seamless handoffs
We designed an AI receptionist to manage intake, safety, and scheduling while keeping conversations consistent. It revealed how reliability under pressure depends on empathy balanced with clear transitions.
Phase 2
Refining Communication for Emotional Balance
Functional reliability held, but reassurance still cracked. We explored how empathy and oversight could sustain confidence under pressure, and how AI might help them coexist without overstepping in care.
Testing how empathy and oversight coexist
We translated these tensions into two probes: Sentiment Cue for empathy, highlighting tone shifts to guide awareness, and Response Support for oversight, surfacing key details to steady staff judgment.
AI receptionist for patients: Practicing safe empathy under oversight
The AI receptionist showed care within limits, clarifying next steps while keeping context visible. It revealed how emotional reassurance depends on clear scope and transparent boundaries.
AI assistant for staff: Balancing efficiency and empathy under pressure
The AI assistant combined Sentiment Cue and Response Support to keep empathy steady amid urgency. By grounding emotion in context, it showed how balanced judgment sustains confidence in complex calls.
Integration
Testing where confidence held or cracked, shaping balanced collaboration
My Impact
Built Vapi prototype to test real dialogues and expose trust cracks.
Translated findings into guardrails for trustworthy human-AI collaboration.
Patient Testing
When Empathy Comforts but Timing Cracks
Simulating patient calls under uncertainty
We recreated urgent calls where patients were anxious and couldn’t recall medication. AI clarified context and relayed details to staff, showing how steady communication can calm uncertainty.
When empathy overreaches and loses credibility
Patients stayed calm when AI remained transparent and carried context forward. When tone felt scripted or safety checks lagged, reassurance turned hollow and care lost credibility.
Staff Testing
When Automation Supports but Judgment Fades
Testing staff walkthrough of urgent call
We ran the same urgent-call scenario with staff. Using AI-shared context and cues, they confirmed details and guided patients toward safe next steps, revealing how oversight can steady routine work.
When automation outpaces judgment
Staff stayed confident when AI handled routine tasks and passed clear context. But when emotion scores lacked clarity or edits locked out discretion, confidence thinned and judgment lost ground.
Synthesis
From Scattered Tensions to Shared Guardrails
Testing revealed consistent patterns of balance. Patients valued warmth that stayed grounded, while staff relied on clarity and control. Together these insights formed four guardrails for confident collaboration.
Transparency
Patients shouldn’t wonder who they’re speaking to; staff need to see how AI decides.
Continuity
Patients shouldn’t feel dropped; staff need smooth carry-over across tasks.
Resonance
Patients expect warmth that feels caring; staff need cues that keep empathy real.
Accountability
Patients deserve accountable care; staff need control to review and override.
Trust Framework
Weaving four principles into one model aligning patients, staff, and AI
My Impact
Synthesized guardrails and tensions into a framework of trust.
Trust Framework
A System for Sustaining Confident Collaboration
Assurance in care isn’t built by single rules, but by coordination. This framework weaves three mechanisms: Flow, Layer, and Balance that keep clarity, empathy, and authority aligned across people and AI.
Mechanisms
Scaling Assurance Across People and AI
Flow: Connecting clarity across handoffs
Trustworthiness held when information and emotion moved as one. Flow captured that precision in handoff, helping patients feel remembered and staff step in smoothly when context changed.
Layer: Attuning support to shifting needs
Assurance deepened when assistance flexed with need. Under stress, warmth surfaced; when focus returned, guidance eased. Layer let care breathe: deep when needed, light when steady.
Balance: Holding empathy and authority in view
Care felt steady only when empathy and authority stayed visible together. When tone grew too personal, trust thinned; when human control faded, judgment slipped. Balance kept both in sight.
Final Concept
Translating Flow, Layer, and Balance into care that keeps patients and staff reassured
My Impact
Refined patient–AI-staff handoff flows from evaluation insights to strengthen reassurance.
Goal
Two Roles, One Continuous Flow
Patients repeated details while staff re-verified without context. The AI receptionist and assistant share one flow, keeping clarity and empathy connected end to end.
AI Receptionist
Starting Flow with Clarity and Care
Patients felt lost repeating symptoms to different people. The AI receptionist recalls context and explains next steps, helping first contact feel clear and reassuring.
Call Assistant
Flow, Layer, and Balance in Action
Before the call: Starting with context through Flow
Staff once started calls blind. AI now shares patient history and response support upfront, cutting repetition and setting a smoother, more confident start.
During the call: Maintaining rhythm through Layer
Under stress, staff missed empathetic tone. AI senses shifts and suggests pacing, keeping empathy measured and care steady.
Sentiment cue
AI flags subtle emotion changes and phrasing gaps, guiding staff to sound genuine without losing clarity.
Task automation
AI completes routines such as scheduling mid-call, keeping patients informed and conversations unbroken.
After the call: Closing with confidence through Balance
Follow-ups often slipped between systems. AI drafts notes for review and override, keeping oversight intact and care accountable.
Outcome
Translating care insights into a trust framework for future human-AI collaboration
For users
Clarity for Patients, Preparedness for Staff
Patients: One story, clear next step
No more repeating themselves. The AI carried context across calls so each started with clarity and ended with calm about what comes next.
“I didn’t have to repeat myself… what I asked for has already been communicated. And I’m good to go.”
- Patient 2
Staff: Prepared calls, human voice
Staff began calls informed and focused. With light AI guidance, they balanced empathy and control, responding with calm and confidence.
“If an AI has already gone through the patient’s emotional state, … I can ease in and make them feel safe.”
– Staff 1
For Hiya
Adoption Guardrails Beyond Healthcare
Probes showed that quick fixes cracked trust instead of building it. I reframed the learnings as adoption guardrails, baseline conditions for trust. Similar gaps appeared in finance, utilities, and government, showing how these conditions scale beyond clinics.
Reflections
From framing clarity to grounding reality, each insight defines what to test next
Learnings
Designing for Trust in Complex Systems
Design leadership in ambiguity
I guided the team through 2030 envisioning and tension mapping across patient and staff journeys. It taught me to lead with frameworks so the team could see direction when the path was open.
Mapping the forces of trust
Mapping tensions across patients, staff, Hiya, and experts showed how different needs pull trust apart. Patients sought reassurance, staff needed coherence, and experts split on empathy. Seeing these pulls revealed where to design and balance.
AI tools as exploration partners
Using Vapi and ChatGPT, I turned tensions into quick probes. AI boosted speed and perspective, helping me explore wider and see where human judgment still matters most.
Grounding vision in reality
Exploring a HIPAA-lite path showed that flexibility without responsibility is fragile. Even basic identifiers required compliance, reframing regulation as credibility, the base where system trust begins.
AfterThoughts
Next Experiments to Deepen the Guardrails
If the project continued, we’d prototype key scenarios to test how each guardrail holds in real interactions.
Sentiment support in real time
Test if intent cues and emotion trends lead to more empathetic responses than raw scores.
Contextual handoff from EHR data
See how much auto-generated context is enough before staff return to records.
Patient support during wait time
Explore which AI-led wait chats calm patients without extra work for staff.
Conclusion
A Small Win in Hiya’s 2030 Domain Exploration
Through a series of research presentations and one alignment workshop with Hiya’s cross-functional teams and leadership, we defined how trust should evolve across patient-AI and staff-AI flows, giving Hiya’s 2030 vision a tangible frame for credible, human-centered care.
