Verisk brings insurance analytics into Anthropic’s Claude via MCP connectors

Verisk pushed its insurance analytics closer to the daily workflow on May 5, 2026, making its data available inside Anthropic’s Claude family through standardized Model Context Protocol connectors. For underwriters and claims teams, that matters because it shifts Verisk’s analytics out of isolated dashboards and into a conversational layer where users can ask for what they need in natural language, then get governed results back without bouncing across systems.

The first connectors include Verisk Underwriting Intelligence, built on ISO Indications, and Verisk XactRestore for restoration use cases. Verisk said Underwriting Intelligence gives insurers conversational access to loss cost trends, experience insights, and filing signals from Insurance Services Office, a Verisk business. The company also said the connectors were designed to support controlled access and compliance needs inside its established data-governance framework, which is the part that will decide whether this feels useful or reckless in a production environment.

Anthropic’s MCP pitch gives the move its technical shape. Anthropic describes MCP as an open standard for connecting AI assistants to the systems where data lives, and on the same day it said it was expanding its financial-services and insurance ecosystem with ten ready-to-run agent templates, new connectors, and an MCP app for financial services and insurance organizations. That combination lowers one of the ugliest costs in enterprise AI, the hand-built integration work that usually turns every new model into a months-long plumbing project.

The practical upside is easy to see. Carriers do not usually lose time because they have no data. They lose time because someone has to find it, interpret it, validate it, and then copy it into the next system. A standardized connector architecture gives Claude a cleaner path into underwriting and claims intelligence, which could make internal copilots more than a demo and turn them into working tools for querying exposure, checking filing signals, or pulling restoration context on demand. The risk, of course, is hallucination if teams treat a model’s answer like a fact pattern instead of a starting point, which is why Verisk’s emphasis on explainability and human accountability matters as much as the interface.

Verisk said its analytics are already used by U.S. P&C insurers, including the top 100 carriers, along with global insurers, reinsurers, and brokers. The company also said it has embedded AI across the insurance ecosystem for more than two decades and has deployed about 40 agentic and generative AI solutions, while its 2025 annual report says more than 35 AI-powered projects and solutions are in use today. That makes the Claude integration look less like a pilot and more like a distribution decision for a large installed base that already trusts Verisk’s data to shape underwriting and claims decisions.