Generative AI in Underwriting: A Practical Guide

Generative AI is reshaping underwriting. It speeds up processes, aids decisions, and flags risks. But this is new territory. Mistakes can happen. Each use teaches us how to get better. Here are some ideas for using AI in underwriting without losing control.

1. Automate Data Inputs

Underwriters spend hours pulling data. AI can handle this.

Example: Loss runs come in many formats, but a well-designed GenAI tool can process them efficiently.

Bottom Line: Automate the grunt work. Let underwriters focus on decisions and communication.

2. Aid Classification

Small businesses often misclassify themselves. AI helps fix that. Business owners describe what they do, and AI refines the details.

Example:“We’re a handyman business.”AI follows up:

• “What specific work do you do?”

• “Do you keep separate accounting for different activities?”

Key Idea: Better questions mean better classifications. Better classifications mean fewer coverage gaps.

3. Surface Hidden Risks

AI can flag risks from public data, websites, social media, and unread loss control reports.

Example 1: A contractor’s website shows workers on rooftops without safety gear. AI catches this and flags it for review. Example 2: Driving radius is buried on page 34 of a report. It exceeds guidelines but would’ve gone unnoticed without AI.

Impact: Fewer surprises. Faster fixes.

4. Automate Decisions

Carriers already use rules engines for automated quotes. AI enhances these systems by feeding accurate data, justifying decisions, and flagging ambiguities.

Example:“We quoted this based on your input. But your site mentions tree services, which may require a coverage change. Please confirm.”

Key Rule: Decisions must tie to real data. Always provide an escalation path.

Keys to Success

• Base actions on facts, not AI guesses. Example: “Your website says X; your form says Y.”

• Start slow and accelerate. Watch outputs and refine as you go.

• Build a feedback loop. Each interaction improves the system.

• Know your risks. State rules vary. Stay compliant.

• Keep humans in the loop. Machines assist; people decide.

The Bottom Line

AI is a tool, not a decision-maker. Use it to automate, flag risks, and speed up tasks. Start cautious. Learn from every case. Build confidence. Done right, AI will make underwriting smarter and free human capacity for growth and profitability.