← Back to work

Helport: AI Copilot for Live Calls

An AI co-pilot that guides health insurance agents through live calls with real-time, policy-aware answers, built-in compliance safeguards, and automated call summaries.

RoleProduct Experience Designer
ForU.S. health insurance agents and members
Team4 — Software Eng, Backend Eng, Business Strategist, Product Designer
TimelinePLACEHOLDER — confirm with Joyce
hero screenshot — Helport
01 — Research

Why We Built This

58%

of insured adults report at least one issue with their health insurance coverage every year

1 in 6

say they were unable to receive recommended care due to insurance hurdles

~30%

annual attrition among the ~70,000 customer service reps at health and medical insurance carriers

Health insurance support in the U.S. is breaking down at scale — for the members trying to use their coverage and the agents trying to help them.

The Gap

U.S. health insurers collectively cover around 300 million Americans, and the industry is already signaling it's ready for AI — CVS Health kicked off a multiyear $2B cost-cutting effort, and UnitedHealth's CTO has said AI investment is core to fixing a health care system that needs it.

02 — Pain Points

Three ways it breaks down

Member Confusion

Members can't tell what's covered, get inconsistent answers, and face delays that affect their health.

Agent Overload

Agents juggle fragmented systems and complex policies, driving long calls, escalations, and compliance risk.

Systemic Inefficiency

Outdated tools drive slow, error-prone support: repeated calls, higher costs, preventable violations.

03 — Problem Statement

The problem

A frontline health insurance agent taking back-to-back calls needs to give an accurate, plan-specific answer while the member is still on the line, because the answer is scattered across portals and policy PDFs that vary by campus, and getting it wrong means a compliance violation or a delay in someone's care.

04 — Solutions

Built to fix it

We turn overwhelmed insurance calls into confident conversations — cutting through confusion with instant answers, clear guidance, and seamless follow-up.

#1

Realtime Policy-Aware Guidance

Empowers agents with instant, plan-specific answers triggered by live call transcription.

#2

Streamlined, Scripted Workflows

Uses guided speech navigation to walk agents through complex, multi-step processes.

#3

Built-In Compliance Safeguards

Detects and flags HIPAA/CMS risks during live calls, providing real-time corrective prompts to avoid legal violations and protect star ratings.

#4

Automated Call Summaries

Generates post-call summaries outlining key issues, coverage info, and next steps.

05 — User Personas

Both sides of the call

Marcus
Jasmine
06 — Target Market

Who we're built for

01Start Here (already built)

Self-funded commercial and student plans

  • UC SHIP is self-funded by the University of California, with medical claims administered by Anthem Blue Cross
  • Anthem is who runs the member services line, so Anthem is who buys
02Expand To

Contact center BPOs

  • Alorica and Teleperformance staff outsourced insurance support and are measured directly on handle time
  • Shortest procurement cycle of any segment
03Then

Medicare Advantage carriers

  • Plans like SCAN Health Plan, a Long Beach nonprofit MA plan
  • Value here includes CMS Star Ratings and CAHPS, which commercial plans don't have
Market Size
  • Roughly 70,000 customer service reps work for health and medical insurance carriers (BLS)
  • At $130 per seat per month, about $109M annually, before counting BPO agents
Why Now
  • Roughly 30% annual agent attrition means constant retraining
  • Margin pressure across the sector is pushing spend toward cost per call
07 — Service Blueprint

Inside a call

Human
Layer
Co-Pilot
Layer
Greeting & HIPAA check
Speech navigation
Advances the script as the call moves
Reason for the call
Intelligent labeling
Identifies who's calling and why
Discuss & guide
Knowledge base
Pulls the right plan's answer
Wrap-up & next steps
Call summaries
Writes the wrap-up note

Demoed against UC SHIP, the University of California student health plan — a natural fit since 58% of insured adults report a coverage issue every year, and international students hit that confusion without a support system they already know.

08 — Design Process

Design Process

Iteration 1
Iteration 1Mid-Fi

Weak visual hierarchy

Scripts, messages, workflow steps, and summaries competed for attention.

High information density

Nested sections and long text blocks increased the amount of information agents had to process during a call.

Unclear AI guidance

AI recommendations blended into the conversation, making them harder to distinguish from customer and agent messages.

Disconnected call controls

Recording relied on a separate modal, interrupting the live workflow.

Buried next steps

Agents could see the workflow, but completed and pending actions were not immediately clear.

Iteration 2
Iteration 2 - Design Outcome

Clear AI guidance

Separated AI recommendations from agent and customer dialogue.

Actionable task tracking

Simplified workflows into clear, easy-to-follow steps.

Integrated call controls

Kept essential call actions accessible without interrupting the workflow.

Stronger visual hierarchy

Reduced distractions and made it easier to identify what to say, what to do, and what comes next.

Final Prototype
01 · Dynamic Task Guidance

Stay on track throughout the call

A real-time task list adapts to the call purpose and conversation, automatically updating completed and remaining actions so agents always know what to do next.

02 · Automated Member Verification

Verify members without manual lookup

The system identifies the member using their ID and existing records, then surfaces relevant profile and plan information for verification, reducing repetitive searches and saving time at the start of each call.

03 · Real-Time Knowledge Assistance

Get the right information at the right moment

When a member asks a specific question, AI retrieves relevant information from the knowledge base and presents it directly in the conversation, helping agents respond accurately without memorizing policies or searching through documentation.

04 · Automated Post-Call Summary

Turn every conversation into actionable documentation

After each call, AI automatically organizes the interaction into a call summary, key notes, follow-up actions, transcript, and recording, reducing manual documentation and making the conversation easier to review later.

09 — Competitor Analysis

Where Helport fits

Hyro
Uniphore
Verint
Use Case

Customer self-service & call automation

Real-time agent assistance

Enterprise CX automation

Pros +
  • Call automation & smart routing
  • Self-service & SMS deflection
  • Healthcare-specific workflows
  • Real-time guidance
  • Knowledge assist & next-best actions
  • Automated summaries
  • Conversation intelligence
  • Broad CX platform
  • Agent copilots
  • QA, analytics & WFM
  • Enterprise integrations
Cons −
  • Primarily customer-facing
  • Optimized for deflection/routing
  • Broad enterprise platform
  • More capabilities than a focused workflow requires
  • Very broad platform
  • Focus extends beyond individual agent workflow

Perceptual Map of Service Positioning

Agent-facing assistanceCustomer-facing automationFocused solutionBroad CX platform
Hyro
Hyro
Uniphore
Uniphore
Verint
Verint
Helport
Helport

Helport AI fills the gap with focused real-time agent guidance, compliance support, and workflow automation, making it a strong fit for complex, policy-driven interactions like the UC SHIP demo.

10 — Pricing Model

Who pays, and how

What they pay
$130 per agent, per month

Core platform plus compliance

Core$100 per seat
Compliance$30 per seat
Incentive tiers+$10 to $20
Minimum12 months
Pilot90 days, 50 seats
What they get back
Over 5x return

One customer profile: 500 agents, 700K calls

Cost$65K per month
Saved$343K per month
AlsoHIPAA exposure avoided
11 — Quantitative Outcomes

The numbers

10–20%

drop in average handle time

Saves up to 2,200 labor hours/month in a 100K-call operation

10–15%

increase in first contact resolution

Fewer callbacks, escalations, and member frustration

$4.90

industry cost per call

Even small time savings translate into hundreds of thousands in annual savings

$2.13M

potential reduction in HIPAA violation fees

Real-time QA and intelligent labeling help prevent costly HIPAA/CMS violations

Next case study
Poopidex: AI-Powered Scat Identification