Announcing Unified Risk Reasoning, a revolutionary leap forward giving brokers, insurers and reinsurers a team of agents, grounded in their view of risk, to validate, refine and finalize risk information ahead of decisioning, exactly as a seasoned expert would.
Until now, automating risk workflows has been characterized as solving data collection. In reality it’s much more; it requires both data collection and an active reasoning process to review, choose, and finalize the ultimate fields on which a risk decision is made.
To action both new business submissions and claims, insurers must understand the context of the case, identify the data required, review contents of the submitted documents, locate the outstanding data from appropriate internal and external sources, then refine and finalize each field. This process consists of a complex series of steps, enacted on a large volume of data. Historically, risk automation workflows have prioritised solving for data collection, lacked integrated reasoning layers and resulted in a persisting need for manual workflows.
Insurance professionals don't build confidence by looking at fields in isolation, they interrogate all of the references to a field using many different submission, internal and external sources. As they connect these fragments, a coherent profile of the risk begins to emerge. An ambiguous entry for one field becomes clear when it's supported by three others.The final value for a field is often a composite which comes from reviewing a variety of candidates.
For example, when underwriting EPL risks, a key factor to consider is employee relations. There may be two conflicting candidates for a single employee count field; the proposal form may state '4,800 employees', while the broker flags the mismatch in the email, reporting 'heads up that the employee figure is incorrect in the form, the latest headcount is 5,300 employees’. Active reasoning factors both sources, and amends the field to state the updated number provided by the broker.
The Cytora Platform has become progressively more powerful, scalable and explainable enabling insurers to materialise proven benefits across multiple workflows and lines of business. Now, Unified Risk Reasoning applies the fully controllable reasoning layer to the entire risk, realizing the original vision of true risk workflow automation: reviewing all of the underlying sources, selecting the most appropriate value, underpinned fully explainable by Chain of Thought, enabling downstream decisioning to take place with full context on how fields have been digitized across multiple sources.
Introducing The Cytora Platform: Unified Risk Reasoning, a proactive reasoning layer that thinks and acts like a human.
Multi source reasoning
The Cytora platform enables schema fields to be fulfilled from different combinations of sources (submission, external, internal, web research) to create composite fields, flexibly defined by configuration users in natural language. Users have full control at a field level to determine the breadth of sources, and can decide to sequence multi-source research, and even prioritize specific websites including company websites. Fields can reference data sources (e.g. web information) to verify information from another (e.g. the submission) providing improved reliability and control. This enables insurers to expand how they author their view of risk to include not just the subject of the field but also the mode of fulfillment across any combination of sources.
Comprehensive data library
Risk submissions almost never contain all of the data required to decision those risks. To be made decisionable, underwriters and claims handlers must often expend precious time manually gathering and rekeying data points from external sources.
Cytora’s network of 70+ data partners is available to add with a few clicks from the configuration menu within the Digitization Center. Configuration users can add new external data sources to schemas without any technical resource requirements for frictionless testing, implementation and scaling. This means the platform can automatically generate required data points from external data sources, requiring no touches from underwriters or claims handlers.
Agentic web research
Structured and defined external data sources are one thing, but what if the required data isn’t available in such a format or from a data provider? Underwriting, Operations and Claims teams often lose time searching the web for the data required to complete risks.
For example, when evaluating the types of activity the insured undertakes there can often be limited information in the actual submitted proposal documents and so underwriters will undertake online research to further enrich their understanding of the scope of the risk. This research provides additional colour to the submitted data and helps underwriters make better assessments when categorising the company into a hazard or risk profile
The Cytora platform can provide these additional insights automatically to reduce manual workflows. Composite fields can be configured whereby fields are fulfilled through a combination of submission and web research. Configuration users can create fields where one source (e.g. web information) is used to verify information in the submission.
Enrichment chains
Many enrichment sources carry data dependencies; sometimes, data is required to gather more data. For example, to quote for a property new business request, property peril data is required to evaluate the property exposure. This requires the extraction of the property addresses, the geolocation of each property using an external data provider, and enrichment with external peril score data from a different data provider.
Previously clients would have had to configure multi-step workflows that introduce complexity, friction and manual steps to the process.
Now, the Cytora Platform introduces automated enrichment chains, enabling the output of one enricher to act as the input for another enricher. The configuration of the sequence of enrichers is handled automatically by the Cytora Platform as it understands the relationships between required data points, their sources and, by extension, the order in which data must be gathered to automatically and completely fulfil risk data requirements in one shot.
This capability unlocks advanced data workflows, allowing you to build richer and more comprehensive profiles by progressively enriching data through multiple external sources.
Using the previous property risk example, this configuration is now performed in the platform without having to orchestrate the data sources. The platform automates this entire sequence: it first validates the addresses, resolves them to their longitude and latitude, and then uses those coordinates to enrich the submission with the required peril scores.
Transparent Chain of Thought reasoning
Chain of Thought is summarized in natural language, presenting key details and reasoning to underwriters and claims handlers in the Cytora Platform’s output, making risks easily understandable and rapidly actionable by your experts. Connecting Unified Risk Reasoning with the human elements of the value chain creates a seamless transition between AI agents and commercial insurance workflows.
This feature is also now available in the human review UI and teach interface at a field level, allowing configuration users and exceptions handlers to understand and connect with Cytora’s agentic reasoning.
The now-legacy characterization of decision-ready risks was simply processing risk submissions and fetching risk data according to tightly instructed parameters. The Cytora Platform’s Unified Risk Reasoning evolves the reality of decision-ready risks, understanding and autonomously reasoning to make risks more complete than they have ever been, and realizing the pre-requisites of driving downstream workflow automation.
Cytora Unified Risk Reasoning is available now to new and existing customers. Book a demo to see Cytora in action on your use cases, lines of business and transaction types >
