Can AI Transform Insurance Data Management?

Moody's introduced Risk Data Refinery on May 5, 2026, an agentic application on its Intelligent Risk Platform designed to automate one of the most labor-intensive processes in property insurance underwriting, converting raw submission data into structured, analytics-ready formats.

Insurance slips and Statements of Values move through underwriting, catastrophe modeling, and portfolio management systems across multiple counterparties like brokers, underwriters, and reinsurance partners.

Each transition requires data extraction, validation, and enrichment that has traditionally required large internal expert teams or outsourced operations working manually or in semi-automated workflows. Risk Data Refinery replaces that process with AI agents that complete the same conversions in seconds to minutes, according to the press release.

"Underwriters know that precision is vital; get an address slightly wrong, miss a contract term, ignore an important exposure attribute, and you could find yourself creating expensive mistakes further down the line," wrote Cihan Biyikoglu, Managing Director of Product and Technology at Moody's, and Alok Kumar, Managing Director of Analytical Services at Moody's, in the launch blog post.

What the Application Does

Risk Data Refinery accepts insurance submissions in Microsoft Excel, text, CSV, and PDF formats across multiple languages.

Once uploaded, a set of specialized AI agents, including an address agent, slip agent, SoV agent, data enrichment agent, and policy coding agent, extract structured details from the submission and convert them into Exposure Data Modules, Moody's structured data container holding detailed exposure, financial, and contract information ready for property underwriting and modeling workflows.

According to the press release, First-pass interpretation accuracy is approximately 95%. With an expert reviewing outputs in supervised mode, the system can reach near 100% accuracy within minutes. Moody's early testing suggests that in most cases supervision will not be required.

The accuracy stems from the training data behind the AI agents. Moody's Analytical Services team processes more than 50,000 complex data submissions each year across all format types and has done so for more than 15 years.

Risk Data Refinery's AI agents were trained on that dataset, incorporating edge cases that Moody's describes as absent from most competing solutions

Integration and Availability

Risk Data Refinery integrates across Moody's existing platform ecosystem, available as a standalone application, within Risk Modeler, UnderwriteIQ, and ExposureIQ, via API, or through a Model Context Protocol server.

The MCP integration reflects a broader pattern of insurance data infrastructure providers making their platforms accessible through conversational AI interfaces, consistent with the direction Verisk took when it embedded its analytics into Claude through MCP connectors on May 5, 2026.

The application is available now as a preview on the Moody's Intelligent Risk Platform under Risk Labs, which Moody's describes as a safe experimentation environment isolated from live production data. Existing clients can request access through their Moody's representative. General availability is expected shortly.