Making complex insurance feel human.
Guiding customers through coverage choices with clear AI assistance, useful next steps, and a path to human support.
Project at a glance
I unified fragmented insurance journeys around customer intent—helping people navigate complex decisions, understand recommendations, and reach human support when needed.
My contribution & collaboration
- I set the AI experience direction and aligned partners across Design, Research, Data Science, Engineering, Product, Content, and AI.
- I led experience strategy across 12+ journeys: SmartQuote, quote review/purchase, service, registration/onboarding, claims/FNOL, and policy/vehicle management.
- I directed research, prototypes, and experimentation, built the Digital Assistance Design Kit, and implemented the Level 1–3 AI Maturity Model with Product and Engineering.
A shared model for useful AI.
A shared path from Level 1 to Level 3.
Implemented with Product + EngineeringThe experience anticipates a relevant need and offers a next step.
Relevant help before another question is needed, with a choice about what happens next.
- Signals that inform assistance
- Proactive suggestions
- A more connected companion experience
- 01
Invite in context
Evaluate entry points, CTA language, suggestion chips, and visual treatments within the customer’s insurance journey.
- 02
Guide a useful next step
Use contextual assistance, clear explanations, and hybrid conversational/traditional interactions to support complex decisions and tasks.
- 03
Make trust actionable
Define AI identity, transparency, error recovery, and human escalation as part of the interaction—not as an afterthought.
Research & opportunity
I embedded research in product strategy, partnering with researchers, data scientists, Product, and Engineering to evaluate whether AI felt useful, understandable, trustworthy, and appropriate for the State Farm brand.
Research artifact
Research questions & methods
Compare conversational and traditional flows, and widget versus full-page assistance, against the customer’s task.
Test entry points, CTA language, suggestion chips, proactive assistance, and visual treatments to understand how people discover and use AI.
Evaluate AI identity, explanations, and human escalation for trust, brand fit, and customer control.
Methods included usability studies, A/B experimentation, prototype validation, and AI trust/brand studies. Specific findings and post-test performance results are not supplied.
Where customers needed more support
Onboarding and service interactions helped frame where clearer guidance and a smoother path to human help could make a difference.
27,700 onboarding & service interactions
Slightly more customers completed without escalation than required escalation. Select a bar to explore.
Nearly half of analyzed interactions required additional support. The design opportunity is to reduce avoidable friction while keeping a clear path to human help.
Reduce unnecessary effort by recognizing customer intent earlier, adapting guidance to context, and proactively surfacing the next-best action. Keep the transition to human support clear and seamless when additional help is needed.
Explore the From → To journey
From friction to purposeful assistance
Select a stage to see the design opportunity.
Recognize the customer’s goal early so the experience can respond to the task rather than requiring a long sequence of steps.
Research coverage & data context
- Entry points
- Conversational vs. traditional UI
- Proactive assistance
- Trust
- Branding
- AI identity
- Human handoff
- Personalization
The interaction counts frame where assistance may help. Dataset dates and measurement definitions were not provided; these are not outcomes from the prototype.
Design around the customer’s task.
Try the shopping experience.
Match assistance to the customer’s task.
Designed full-page AI, widgets, side panels, embedded assistance, and mobile experiences, with transitions between conversational and traditional interfaces.
Give AI a shared interaction foundation.
Built and governed the Digital Assistance Design Kit: prompt and input behavior, quick replies, AI-generated cards, explainability, human handoff, error recovery, trust, and responsive desktop/mobile patterns.
What this work enabled.
- EstablishedShared AI design foundation
- Digital Assistance Design Kit, interaction principles, and a Level 1–3 maturity model aligned with Product and Engineering.
- Next directionMore proactive assistance
- Predictive guidance and agentic task completion remained longer-term opportunities.
Agentic experiences were a longer-term direction. Program scale describes reach and responsibility; measured conversion, service-resolution, or customer-satisfaction results are not provided.