AI Agents

Context

Built for insurance within the Reinsurance coordination stack, this agent is used whenever a new or amended risk needs to be assessed against reinsurance treaties—at new business, renewal, endorsement, or portfolio true-up. It sits early in the reinsurance flow so eligibility is clear before placements or notifications, and later when endorsements change key attributes (sum insured, peril, territory, or terms).

What it does

The agent reads the policy record and relevant treaty set, then determines whether the risk is cedable and under which agreement. It parses treaty wording to identify scope (lines of business, perils, territories), attachment points, limits, deductibles, exclusions, hours clauses, risk-attaching vs. loss-occurring bases, and special conditions. It matches these to the policy’s characteristics—class, peril, geography, effective dates, sums insured, sublimits, and endorsements—and outputs a clear eligibility decision with the treaty reference, clause citations, and any conditions to satisfy (e.g., exclusions triggered or facultative required). Each conclusion is linked to the exact treaty passages and policy fields for quick verification.

Core AI functions

Document classification and clause extraction for treaty and slip wording; normalization of terms, limits, and dates; policy-to-treaty rule evaluation that maps coverage scope, basis (risk attaching / loss occurring), and attachment structures; exception surfacing for edge cases (e.g., mixed perils, co-insurance, layered towers); and reason-code generation that explains why the risk is eligible or not, with clause-level references. Version awareness ensures the agent uses the correct treaty year and endorsement edition.

Problem solved

Manually checking eligibility across multiple treaties is slow and error-prone. Mis-ceded or un-ceded risks lead to rework, delayed notifications, or disputed recoveries. Endorsements late in the cycle can quietly push a risk outside scope without being noticed.

Business impact

Faster, cleaner cessions with fewer errors; treaty capacity is used correctly; recoveries are protected with clause-level evidence; and downstream placement and accounting move on time. Teams gain confidence that eligibility is consistent across portfolios and audit-ready.

Integration and adjacent use cases

Integration is moderate: the agent reads policy/exposure data from PAS or your data warehouse and treaty documents from your repository or DMS; it writes eligibility decisions, references, and conditions into the reinsurance workbench or orchestration layer—no changes to cores required.

Common combinations in this stack:

• Retention limit validator (to confirm retained amounts and attachments),

• Reinsurance treaty cross-referencer (to align policy and claim data with the right clauses),

• Claim sharing agreement trigger agent (to initiate sharing when applicable), and

• Reinsurer notification & documentation assistant (to assemble and send the required packs).