high72
Draft and Send Claim Correspondence
Communication
Microsoft Copilot for Microsoft 365
AI tools like Harvey (legal/insurance-specialized LLM) or Microsoft Copilot can draft the bulk of structured claim correspondence — reservation of rights letters, coverage decision notices, settlement position letters — using claim data, jurisdiction-specific language templates, and regulatory requirements, saving 65-75% of drafting time.
medium62
Document Claim File Activity
Administrative
Guidewire ClaimCenter
AI tools like Claude or GPT-4 can draft structured diary notes, coverage analysis summaries, reserve rationale narratives, and action plan templates from claim data inputs (adjuster notes, medical records, loss runs) — handling roughly 55-60% of the writing and formatting work.
medium42
Review and Evaluate New Claims
Research
Guidewire ClaimCenter
AI tools like Gradient AI or Verisk's claims analytics platforms can accelerate document ingestion, flag coverage gaps, extract key data from medical records and incident reports, and surface comparable claim benchmarks — handling roughly 40-50% of the research and data-gathering execution time.
medium42
Complete Reinsurance Reporting Documentation
Reporting
Guidewire DataHub
AI tools can meaningfully accelerate the identification of claims exceeding reinsurance thresholds by querying structured claims data, auto-populating loss summaries and reserve data into report templates, and flagging treaty-specific reporting deadlines — potentially handling 40-50% of execution time.
low38
Set and Adjust Claim Reserves
Data Processing
Gradient AI's claims intelligence platform can ingest claim facts and medical data to generate statistically-derived reserve range recommendations and flag severity predictors, cutting the data-gathering and benchmarking phase by ~35% while the examiner reviews outputs and applies jurisdiction-specific and program-specific judgment before finalizing the reserve.
Reserve-setting requires synthesizing highly contextual inputs — medical prognosis trajectories, jurisdiction-specific indemnity benchmarks, litigation exposure, and employer-specific loss-sensitive program implications — that current AI tools can partially surface but cannot reliably weight or integrate into a defensible reserve figure.
low28
Coordinate Medical and Return-to-Work
Communication
Microsoft Copilot (integrated into Outlook/Teams) can draft structured follow-up emails to physicians and employers, auto-summarize medical documentation, and generate RTW status update notes — reducing administrative wrap-up time by 20-30% while the examiner reviews and sends each communication.
This task is fundamentally driven by live multi-party coordination — phone calls and negotiations with physicians, nurse case managers, and employers — where relationship dynamics, medical judgment interpretation, and real-time persuasion determine outcomes.
low22
Manage Litigation and Legal Strategy
Strategic
Harvey (AI legal assistant) can summarize legal filings, draft response memos to defense counsel, and flag anomalous billing line items in invoices, reducing document review prep time by 30-40% — though the examiner still makes all strategic and settlement decisions.
This task is dominated by real-time human judgment activities — directing litigation strategy, making live settlement authority decisions at mediations, and negotiating with opposing counsel — that AI cannot meaningfully execute even in a HITL model.