AI-Assisted Claims Experience
Concept Design
Human-Centered AI Workflow for Insurance Claims
Focus: Agentic AI, insurance UX, transparency, trust
Summary
A concept study
Exploring how agentic AI can make insurance claims more transparent, faster, and more human.
This self-initiated project was inspired by my interview for the Senior Manager, AI UX/UI Design role at an insurance company.
It reimagines the claims experience with an AI-assisted workflow that helps customers understand progress, predict next steps, and build trust through clear, conversational design.
Problem & Opportunity
Filing a claim should be simple, but it rarely is
Problem
Filing a claim can be a stressful and confusing process. Customers struggle to know what to include and how to describe incidents, while adjusters spend time verifying incomplete data.
Pain Points
-
Long, manual data entry
-
Lack of real-time guidance or status clarity
-
Low confidence in whether the claim is complete or accurate
The challenge was to simplify this experience without sacrificing accuracy or control.
The Concept – AI Claims Pilot
Where AI supports, not replaces, the user
AI-Human Collaboration Model
- AI Observes – Detects damage and key details from uploaded photos.
- AI Suggests – Prefills claim type, severity, and likely next steps.
- User Decides – Reviews and edits each recommendation before submitting.
- AI Explains – Shows reasoning (“Based on photo analysis and past cases”).
This model improves efficiency while reinforcing trust and transparency.
Dashboard
Upload Screen
Review Screen
Confirmation
Design Principles
| Principle | Description |
|---|---|
| Transparency | Always show what the AI did and why. Use explainable microcopy like “Suggested by AI based on photo analysis.” |
| Control | The user remains in charge. All AI suggestions are editable or dismissible. |
| Guidance | Replace complexity with contextual tips and progress cues instead of long forms. |
| Trust | Use calm visuals, predictable interactions, and confidence indicators to make AI behavior feel dependable and human-aligned. |
Experience Impact
For Users
-
Turns a stressful, manual process into a guided, confident experience.
-
Reduces errors and incomplete submissions.
-
Gives transparency into claim progress and timelines.
For the Business
-
Fewer incomplete claims → faster resolution cycles.
-
Reduced manual verification workload.
-
Establishes a scalable foundation for future agentic AI workflows.
Metrics to Track
While this concept was not implemented, I defined potential experience and operational metrics to measure the success of an AI-assisted claims workflow.
User Experience Metrics
-
Claim filing time: Average time to complete a claim submission (target: reduce by 40–60%).
-
Error rate: Frequency of incomplete or incorrect submissions (target: reduce by 30%).
-
User confidence score: Collected via post-submission survey or sentiment analysis (“How confident do you feel your claim is complete?”).
-
Transparency rating: % of users who report understanding why the AI made a suggestion.
Operational / Business Metrics
-
Manual verification load: % reduction in adjuster time spent clarifying missing info.
-
Claim resolution time: Average days from submission to first agent action.
-
Reopen rate: % of claims reopened due to missing or incorrect data.
-
Adoption rate: % of users who choose to use the AI-assisted flow versus manual filing.
Reflection
This exploration demonstrates how human-centered design can shape emerging AI technologies into experiences that feel supportive rather than opaque.
It also reflects my approach to leading design in new domains—by rapidly learning, framing problems through empathy, and visualizing strategic possibilities.
MORE CASE STUDIES
Explore Other Work
Leadership stories and design solutions across regulated industries
