Insurtech vendors roll out agentic products

Insurance has spent years trying to bolt generic chatbots onto workflows that still run on PDFs, email chains, and policy wording written like legal code. In the past two weeks, four vendors launched products that skip the general chatbot pitch and go straight at the jobs insurers actually do: underwriting and claims. (datamatics.com, prnewswire.com, instanda.com, finance.yahoo.com) Underwriting is the part where an insurer decides what risk it is taking and what price to charge for it. Claims is the part after something goes wrong, when the insurer has to read the policy, read the facts, and decide what gets paid. (iii.org, naic.org) Both jobs are document-heavy in a very specific way. A human underwriter might read medical records, loss runs, inspection reports, and broker submissions, while a claims lawyer might compare a directors and officers policy against a lawsuit paragraph by paragraph. (datamatics.com, claimence.com) That is why these new products all look less like open-ended assistants and more like narrow operating systems. They ingest files, extract fields, apply rules, surface exceptions, and leave a human with a shorter stack of decisions instead of a blank chat box. (datamatics.com, claimence.com, convr.com) Datamatics said on April 9 that its TruAI Underwriting platform can cut underwriting turnaround time by up to 70 percent, lower costs by up to 50 percent, and improve accuracy by up to 25 percent. The company says the system pulls data from structured and unstructured documents, builds a digital case summary, and keeps a human reviewer in the loop. (datamatics.com, scanx.trade) Claimence went after a different bottleneck on April 8: financial lines claims, starting with directors and officers liability. Its pitch is that a single directors and officers claim can trigger thousands of dollars of outside counsel review, and its software parses policies and claim files to generate coverage analysis and correspondence faster. (finance.yahoo.com, claimence.com) Convr’s March 17 release sits one step earlier in the pipeline, inside commercial insurance underwriting. It embedded a conversational assistant into its underwriting workbench so underwriters can interrogate submission data inside the same system that already handles intake, triage, and risk review. (prnewswire.com, convr.com) INSTANDA’s March 31 launch is aimed at the ugly edge cases that break simpler automation. The company says INSTANDA MAX lets commercial line and non-admitted insurers underwrite tens of thousands of complex assets under one policy in real time, at both portfolio level and individual item level. (instanda.com, insurtechanalyst.com) Put those four launches together and a pattern appears. Vendors are no longer selling “artificial intelligence for insurance” in the abstract; they are selling software for one insurance task, one line of business, one document set, and one approval chain at a time. (datamatics.com, claimence.com, prnewswire.com, instanda.com) That shift fits how insurers buy technology. Carriers usually do not replace the whole core system in one move, but they will fund a tool that removes hours from underwriting a life policy, or cuts legal spend on a directors and officers claim, or helps a commercial underwriter clear a backlog before renewal season. (mckinsey.com, datamatics.com, finance.yahoo.com) The next test is not whether these products can write fluent answers. It is whether they can survive audits, explain why a rule fired, handle messy submissions from brokers, and earn enough trust that an underwriter or claims handler will actually let the machine touch a live file. (naic.org, datamatics.com, instanda.com)