Decerto expands Claims AI as US insurers move to production

Decerto’s June 4 expansion of Claims AI landed at a telling moment for US property and casualty insurers: the conversation has moved from testing AI to putting it into production. In claims, where speed, customer experience and loss control all collide, carriers are looking for automation that can handle triage, document intake, fraud screening and decision support without breaking the flow of an existing operation.

That is why Decerto is positioning Claims AI as a production stack rather than a narrow point tool. The platform is built to connect with major US PAS architectures, financial and accounting systems, document management tools, and external fraud data sources such as NICB and ISO ClaimSearch. It also handles PDFs, mobile photos, scanned paper and handwritten notes, with Decerto saying it can extract data with more than 99% accuracy and move a clean submission through nine decision layers in roughly 90 seconds. The promise is not just faster intake; it is a claim path that can be traced, explained and audited end to end.

That traceability matters because the operational problem is bigger than model performance. The National Association of Insurance Commissioners adopted its Model Bulletin on the Use of Artificial Intelligence Systems by Insurers on December 4, 2023, and the guidance centers on fairness, accountability, transparency, compliance and a safe, secure, fair and robust system. Decerto’s own 2026 guide adds a hard reality check: straight-through processing in claims is still below 10% on average, and nearly 60% of insurers reported no STP at all in claims operations. It also says the bulletin was in effect in 24 US jurisdictions as of August 2025. The pressure now is not simply to automate more, but to automate in a way that survives regulatory scrutiny and still leaves room for exceptions, overrides and human review when the file is messy.

Fraud workflow is one of the clearest examples of where integration counts. Verisk says ClaimSearch is the property and casualty industry’s largest claims database, while NICB describes ISO ClaimSearch as the industry’s All Claims Database, covering property, casualty and automobile claims data. NICB also says its Information Submission service lets member companies report questionable activity even when it is not directly tied to a claim. That kind of shared intelligence is exactly what claims AI vendors are trying to plug into, because suspicious patterns rarely stay inside one carrier’s system.

The financial stakes are large enough to justify the push. Deloitte estimates AI-driven technologies across the claims life cycle could save P&C insurers between $80 billion and $160 billion by 2032. Decerto is also framing speed as retention, saying 87% of policyholders renew based on claims speed, and its own benchmarks put a manual P&C claim at about 70 minutes and roughly $50 in loaded labor versus 5 minutes and about 7 cents automated. The message is clear: claims AI is no longer being sold as a demo. It is being sold as infrastructure, and the carriers buying it will be judged on how well it fits the real machinery of production.