Clearcover
Leveraging AI to create a faster, smarter, and more human claims experience at Clearcover.
Summary
At Clearcover, I led the design of an AI-powered claims intake experience that brought the empathy and expertise of a claims representative into a self-guided, digital format. The goal was to reduce manual review time, improve reporting accuracy, and accelerate payouts.
Policyholder Claims flow used as a reference for our work
Summary
At Clearcover, I led the design of an AI-powered claims intake experience that brought the empathy and expertise of a claims representative into a self-guided, digital format. The goal was to reduce manual review time, improve reporting accuracy, and accelerate payouts.
Policyholder Claims flow used as a reference for our work
Summary
At Clearcover, I led the design of an AI-powered claims intake experience that brought the empathy and expertise of a claims representative into a self-guided, digital format. The goal was to reduce manual review time, improve reporting accuracy, and accelerate payouts.
Policyholder Claims flow used as a reference for our work
Key Results
2 day reduction in claims payout
Reduced Cost per claim
Reduction in Claims related calls and emails
Role & Timeline
Product Designer
2 months
Methods
Visual Design
Prototyping
LLM Training
Interviews
Understanding the problem
Traditional claims intake was slow and inconsistent. Customers often submitted incomplete stories, which forced claims reps to follow up manually by phone or email. This delayed resolutions, increased costs, and added frustration for everyone involved.
We needed a way to collect better, clearer information up front, without losing the human-centered trust that matters during stressful, high-emotion events.
High level look at the steps involved in claims data collection
Understanding the problem
Traditional claims intake was slow and inconsistent. Customers often submitted incomplete stories, which forced claims reps to follow up manually by phone or email. This delayed resolutions, increased costs, and added frustration for everyone involved.
We needed a way to collect better, clearer information up front, without losing the human-centered trust that matters during stressful, high-emotion events.
High level look at the steps involved in claims data collection
Understanding the problem
Traditional claims intake was slow and inconsistent. Customers often submitted incomplete stories, which forced claims reps to follow up manually by phone or email. This delayed resolutions, increased costs, and added frustration for everyone involved.
We needed a way to collect better, clearer information up front, without losing the human-centered trust that matters during stressful, high-emotion events.
High level look at the steps involved in claims data collection
What we learned
I worked with claims reps, customers, and the AI team to explore how people prefer to tell their story during a claim. Research showed customers naturally share more detail when talking with a human, even without being prompted. I wanted to recreate that sense of openness and empathy in a digital flow.
We tested three input methods in unmoderated sessions:
🗣️ Voice narrative (audio recording)
📹 Video narrative (self-recorded video)
⌨️ Manual text entry (open text field)
Entry into our 3rd Party Claims experience.
Recording of the Audio Narrative Prototype
What we learned
I worked with claims reps, customers, and the AI team to explore how people prefer to tell their story during a claim. Research showed customers naturally share more detail when talking with a human, even without being prompted. I wanted to recreate that sense of openness and empathy in a digital flow.
We tested three input methods in unmoderated sessions:
🗣️ Voice narrative (audio recording)
📹 Video narrative (self-recorded video)
⌨️ Manual text entry (open text field)
Entry into our 3rd Party Claims experience.
Recording of the Audio Narrative Prototype
What we learned
I worked with claims reps, customers, and the AI team to explore how people prefer to tell their story during a claim. Research showed customers naturally share more detail when talking with a human, even without being prompted. I wanted to recreate that sense of openness and empathy in a digital flow.
We tested three input methods in unmoderated sessions:
🗣️ Voice narrative (audio recording)
📹 Video narrative (self-recorded video)
⌨️ Manual text entry (open text field)
Entry into our 3rd Party Claims experience.
Recording of the Audio Narrative Prototype
Key results from our research showed:
🥇 Voice recording was preferred overall. Participants noted that it felt more natural and “hands-free,” especially when trying to recall details or describe an event in their own words.
🥈 Text entry came in a close second, praised for being simple and non-intrusive—especially for users in quieter environments or without access to a mic.
🥉 Video was the least preferred. Participants described it as “awkward” and “distracting,” with many saying that seeing themselves on camera made them feel self-conscious and less focused on the event they were trying to describe.
Key results from our research showed:
🥇 Voice recording was preferred overall. Participants noted that it felt more natural and “hands-free,” especially when trying to recall details or describe an event in their own words.
🥈 Text entry came in a close second, praised for being simple and non-intrusive—especially for users in quieter environments or without access to a mic.
🥉 Video was the least preferred. Participants described it as “awkward” and “distracting,” with many saying that seeing themselves on camera made them feel self-conscious and less focused on the event they were trying to describe.
Key results from our research showed:
🥇 Voice recording was preferred overall. Participants noted that it felt more natural and “hands-free,” especially when trying to recall details or describe an event in their own words.
🥈 Text entry came in a close second, praised for being simple and non-intrusive—especially for users in quieter environments or without access to a mic.
🥉 Video was the least preferred. Participants described it as “awkward” and “distracting,” with many saying that seeing themselves on camera made them feel self-conscious and less focused on the event they were trying to describe.
Designing the experience
After reviewing these insights with product, engineering, and legal, we chose a text-based narrative approach. Legal stakeholders flagged compliance risks with voice or video under laws like BIPA and GDPR. A text-first AI experience gave us the benefits of narrative storytelling while protecting user privacy.
Key design choices included:
• Prioritizing natural language inputs over rigid forms
• Clearly indicate progress as data is collected to help set expectations of how much longer the process would take
• Crafting the AI agent’s tone to be empathetic and human-like
• Aligning prompts with real claims language to improve data quality
I created prototypes using conditional logic and conversation-friendly patterns. We tested them to refine tone, clarity, and trust in the assistant.
Designing the experience
After reviewing these insights with product, engineering, and legal, we chose a text-based narrative approach. Legal stakeholders flagged compliance risks with voice or video under laws like BIPA and GDPR. A text-first AI experience gave us the benefits of narrative storytelling while protecting user privacy.
Key design choices included:
• Prioritizing natural language inputs over rigid forms
• Clearly indicate progress as data is collected to help set expectations of how much longer the process would take
• Crafting the AI agent’s tone to be empathetic and human-like
• Aligning prompts with real claims language to improve data quality
I created prototypes using conditional logic and conversation-friendly patterns. We tested them to refine tone, clarity, and trust in the assistant.
Designing the experience
After reviewing these insights with product, engineering, and legal, we chose a text-based narrative approach. Legal stakeholders flagged compliance risks with voice or video under laws like BIPA and GDPR. A text-first AI experience gave us the benefits of narrative storytelling while protecting user privacy.
Key design choices included:
• Prioritizing natural language inputs over rigid forms
• Clearly indicate progress as data is collected to help set expectations of how much longer the process would take
• Crafting the AI agent’s tone to be empathetic and human-like
• Aligning prompts with real claims language to improve data quality
I created prototypes using conditional logic and conversation-friendly patterns. We tested them to refine tone, clarity, and trust in the assistant.
Impact and outcomes
While this experience was still pending rollout, our goals included:
💰 Faster claims payout time
📉 Reduced cost per claim
☎️ Fewer follow-up claims calls and emails
Next Steps
If I continued on this project, I’d focus on analyzing live data after launch to measure impacts for claimants and claims representatives. I’d also look for ways to strengthen our user prompts, balancing empathy with more precise data collection to further improve reporting accuracy.
Next Steps
If I continued on this project, I’d focus on analyzing live data after launch to measure impacts for claimants and claims representatives. I’d also look for ways to strengthen our user prompts, balancing empathy with more precise data collection to further improve reporting accuracy.
Next Steps
If I continued on this project, I’d focus on analyzing live data after launch to measure impacts for claimants and claims representatives. I’d also look for ways to strengthen our user prompts, balancing empathy with more precise data collection to further improve reporting accuracy.
Thanks!
Thank you so much for taking the time to view my work.
Thanks!
Thank you so much for taking the time to view my work.
Thanks!
Thank you so much for taking the time to view my work.