medium62
Communicate Decisions to Agents
Communication
Microsoft Copilot for Microsoft 365
AI tools like Copilot or specialized insurance communication platforms (e.g., Salesforce Einstein, Grammarly Business, or custom LLM-based drafting tools) can generate structured email drafts for declinations, modifications, and resubmission guidance based on templated decision logic — handling roughly 55-60% of execution time.
medium62
Document Underwriting Rationale and Files
Administrative
Microsoft Copilot for Microsoft 365
AI tools can draft structured underwriting rationale, risk notes, and exception justifications from existing data inputs (application details, prior correspondence, risk flags) — handling the bulk of documentation composition.
medium52
Review and Underwrite New Submissions
Data Processing
Majesco Underwriting360
AI tools can significantly accelerate the data-gathering and preliminary risk-scoring phases — ingesting property details, pulling loss history, flagging coverage anomalies, and generating a structured risk summary with a preliminary accept/decline/modify recommendation — handling roughly 50-60% of execution time.
medium52
Price and Quote Individual Risks
Technical
Guidewire's AI-powered rating and analytics platform (or similar insurtech tools like Cytora) can auto-ingest submission data, run exposure scoring, and pre-populate a draft quote with suggested credits/debits, cutting data entry and initial pricing time by 50%+ while the underwriter reviews and adjusts for bespoke risk factors.
AI tools like Guidewire's predictive analytics, Majesco, or embedded rating engines can automate data ingestion, exposure scoring, and initial premium calculation — handling roughly 50-60% of execution time.
medium52
Monitor and Analyze Portfolio Quality
Reporting
Verisk Analytics Portfolio Manager
AI tools can significantly accelerate the data aggregation, dashboard generation, anomaly detection, and preliminary trend identification phases — tools like Tableau with Einstein Analytics, Microsoft Power BI Copilot, or Databricks AI can auto-surface loss trends, concentration risks, and segment underperformance with minimal manual querying.
low32
Coordinate Risk Management Assessments
Research
Harvey AI (or a similar insurance-specialized LLM) can ingest and summarize property inspection reports, appraisals, and exposure documents, then draft structured coverage adjustment recommendations — reducing document review and memo-drafting time by ~30-40% while the underwriter validates findings and approves final actions.
This task is deeply relational and judgment-intensive — it requires coordinating with external vendors and internal risk teams, interpreting complex property inspection and appraisal data for high-net-worth clients, and making nuanced coverage adjustment recommendations.
low28
Educate and Coach Agent Partners
Communication
ChatGPT or Claude can draft customized pre-call briefing documents, agent-specific FAQ sheets on Berkley One's risk appetite, and follow-up email summaries after coaching sessions — reducing prep and follow-up time by ~30-40% while the underwriter leads the actual conversation and approves all outbound content.
This task is fundamentally relationship-driven and requires real-time adaptive communication — explaining nuanced risk appetite, reading agent comprehension, and tailoring coaching to each partner's specific submission patterns and gaps.
low22
Contribute to Underwriting Process Improvements
Strategic
Notion AI or Microsoft Copilot can auto-generate structured meeting summaries, draft feedback memos on guideline updates, and surface workflow inefficiency patterns from prior meeting notes — reducing documentation overhead so the underwriter can focus on the actual deliberation and decision-making.
This task is fundamentally collaborative and relational — it involves live participation in team meetings, real-time judgment on workflow inefficiencies, and nuanced feedback on appetite/guideline changes that require institutional knowledge and stakeholder dynamics.