Verisk Unleashes Claude AI

**Key Takeaways**

• Verisk launched Model Context Protocol (MCP) connectors on May 5, 2026, integrating its proprietary insurance analytics, specifically Verisk Underwriting Intelligence and XactRestore, directly into Anthropic’s Claude AI platform.
• The integration lets insurance and property restoration professionals access regulatory-grade data conversationally, with Verisk estimating savings of hundreds of hours annually in underwriting and up to two hours per claims estimate.
• Verisk’s approach prioritises governed AI and human oversight, drawing on more than two decades of embedding AI in insurance to keep outputs explainable and data access secure.

[Verisk](https://www.verisk.com) just gave underwriters and restoration contractors a direct conversational line into its regulatory-grade insurance data, piping it straight into [Anthropic](https://www.anthropic.com)‘s Claude via new Model Context Protocol (MCP) connectors launched May 5, 2026. The pitch is straightforward: instead of navigating multiple proprietary dashboards to pull loss cost trends or property repair pricing, professionals ask Claude in plain language and get data-backed answers on the spot. For an industry that runs on accurate data and tight compliance, that’s a meaningful shift in how the work actually gets done.

What the Integration Actually Does

Verisk has built two connectors for Claude. The first, Verisk Underwriting Intelligence, pulls from ISO data, including loss cost trends, experience insights and filing signals. The second, XactRestore, gives restoration contractors conversational access to Xactware pricing and estimating data.

Both connectors are designed to sit inside secure enterprise AI environments rather than open-ended consumer AI setups. The MCP framework controls what data Claude can access and how, preserving Verisk’s data governance protocols and keeping outputs tied to authoritative, verifiable sources. Think of it less as “ChatGPT for insurance” and more as a governed query layer over structured, compliance-grade datasets.

Verisk says the Underwriting Intelligence connector could save carriers hundreds of hours per year by consolidating data access and speeding up assessment of indications and emerging patterns. The XactRestore connector, according to the company, may cut estimate preparation time by 30 minutes to two hours per estimate for experienced contractors. Those are Verisk’s own figures, so real-world results will need to bear them out, but the direction of travel is credible given how manual the current process is.

Two Decades of Insurance AI, Now Conversational

Verisk isn’t new to this space. The company claims to have deployed roughly 40 agentic and generative AI solutions across the insurance ecosystem over more than two decades. That history matters here. The MCP connectors aren’t bolting a general-purpose LLM onto unstructured data. They’re giving Claude structured, governed access to datasets that already underpin under