Bringing efficiency and explainability to complex underwriting workflows
With the ambition to pioneer AI for insurance, Sunlight saw an opportunity to turn this operational complexity into guided intelligence.
The goal: design an AI copilot that helps underwriters build more accurate, efficient, and confident coverages while enhancing, not replacing, their expertise.
To achieve this, the experience needed to balance automation and explainability: an assistant capable of surfacing relevant insights, suggesting coverage improvements, and justifying its reasoning transparently.
01. Contextual and trustworthy entry point to AI assistance
Insurance professionals operate in data-heavy, high-stakes contexts. They’re used to deterministic systems, not generative ones.
The assistant’s first interactions needed to:
Build trust through transparency and predictability,
Demonstrate usefulness in the actual underwriting context, and
Avoid the “AI gimmick” effect common in early copilots.
02. Proactive AI suggestions within the current workflow context
For AI assistance to add value, it needed to:
• Integrate seamlessly into the quoting interface, not sit beside it,
• Provide actionable insights without overwhelming the user, and
• Explain its reasoning clearly enough to earn underwriters’ trust in every suggestion.
03. Designing conversational explainability and data exploration
Empower underwriters to understand and validate AI-generated recommendations by combining conversational insight with data visualization — turning Sunny’s intelligence into a transparent, auditable, and interactive experience.
Users needed to understand why those suggestions were made and what data supported them.
The design challenge was to create a unified interaction model.
Sunny helped underwriters complete quotes twice as fast, reduce manual reviews and report higher confidence in AI-driven recommendations.
By combining proactive insights and explainable reasoning, the assistant turned complex underwriting into a guided, trustworthy workflow.
