Beyond the analogue nature of risks submissions, commercial insurers are not in control of the data structure of the submissions they receive which contain heterogeneous fields that do not conform with their own criteria. Evaluating the eligibility of thousands of different field variations has historically required lengthy interpretation by risk professionals whose time is limited by their scarce capacity. Cytora automatically auto-matches heterogeneous field and declaration variations to standardised pre-approved fields based on similarity to unlock streamlined risk decisioning. Customers receive automation lift in lines of business where heterogeneous fields are common. Cytora provides risk interpretation as a core building block of risk digitization workflows.
Cytora enables insurers to auto-match declaration and field variations to their own defined question sets, providing standardization in a context where insurers receive risks submissions with a multitude of different data structures.
Zero shot training
Zero shot tuning enables insurers to configure and tune risk interpretation models with limited training data. Zero shot inference enables insurers to go live with interpretation powered fields in low data environments, reducing upfront costs of deploying new inference models.
Business users have access intuitive approval workflows that provide visibility and control over the training examples injected into back into the risk interpretation layer to drive continuous optimization. Risk professionals can review, approve and reject training examples to provide control over model learning and accelerate performance.
At Markel, we are committed to giving brokers and clients enhanced service levels and Cytora will enable us to create capacity to effectively grow our scale in our key sectors and enable our underwriters to provide further enhanced service to customers and brokers."