AI Powered Loss Run Triage for Underwriting at Scale скачать в хорошем качестве

AI Powered Loss Run Triage for Underwriting at Scale

AI Loss Run Analysis for Underwriting at Scale (with Audit-Ready Outputs) This webinar shows how Kolena uses specialized AI agents to automate loss run analysis for underwriting at scale, reducing manual re-keying and copy/paste across mixed carrier formats (PDFs, spreadsheets, scanned images). The presenter explains why general chat LLM tools fall short for document-heavy workflows (one-document-at-a-time processing, non-normalized outputs, hallucination risk, lack of audit trails), then demonstrates Kolena’s workflow: upload a multi-document package, run multiple extraction prompts in parallel, and produce structured outputs aligned to a chosen template. Key features highlighted include field normalization, calculations and risk assessment beyond simple extraction, citations that link each value back to the exact source location, and a reasoning trace for QA. The session also covers exporting results, editing/adding prompts, a public loss run agent that emails results, zero data retention policies, continuous benchmarking for quality, and support from insurance-domain AI architects during a two-week trial and follow-on adoption period. Linkedin:   / kolena-ai   Twitter:   / kolenaio   YouTube:    / @kolenaai   Website: https://www.kolena.com/loss-run 00:00 Welcome + Agenda: Loss Run Analysis with AI at Scale 00:47 What Kolena Does (and Where It Fits in Insurance) 01:50 Why Loss Run Analysis Is Still a Manual Bottleneck 03:04 Why ChatGPT/Copilot Alone Falls Short for Loss Runs 05:45 The Specialized Agent Workflow: Inputs → Prompts → Outputs 07:42 Live Demo Starts: Inside a Kolena Account & Agents 10:15 Uploading a Mixed Document Package (PDFs, Images, Spreadsheets) 12:20 Prompt Deep Dive: Claim Type Summary + Citations & Reasoning 15:32 Risk Assessment Prompt: Turning Loss Runs into Underwriting Insights 17:21 Customize Prompts + Add New Extractions 18:14 Exporting the Final Output PDF (What You Receive) 24:05 Wrap-Up and Thanks