How OCEAN works

Live agent

The AI layer over OAK's risk & capital data — every AWS component, method, and MCP tool. Deployed in Alvio's AWS (us-west-2).

Architecture at a glance

Intakeform · CSV · API · threat event

EventBridgeevent bus

Step Functionsocean-assess · durable

AgentCore Runtimethe agent

DynamoDBbook + decisions

This appUI · live trace

⌄

Inside the AgentCore Runtime — the agent plans, calls tools, reasons

Bedrock · Claude Haikureasoning + structured output

AgentCore Gateway (MCP)12 tools → engines + live feeds

Memory (STM)multi-turn context

Guardrailsrecommend-only

AgentCore Observability — OpenTelemetry traces across every step (CloudWatch GenAI dashboard)

AWS components

click any to open it in the live console ↗

Amazon EventBridge

Event backbone — RiskSubmissionReceived and CyberThreatObserved events trigger assessment.

AWS Step Functions

Durable, auditable orchestration (ocean-assess): Normalize → AssessAndPersist → EmitCompleted, with retries.

Bedrock AgentCore Runtime

Hosts the agents — ocean_uw (underwriting) and ocean_copilot (analytics assistant). Plans, calls tools, reasons.

AgentCore Gateway (MCP)

Exposes the analytics engines + live feeds as 16 MCP tools the agents call over JWT-authed MCP.

AgentCore Memory (STM)

Per-session multi-turn memory for the copilot — resolves follow-ups against earlier turns.

Bedrock Guardrails

Recommend-only policy enforced as an auditable post-hoc check on every output.

AgentCore Observability

OpenTelemetry traces of every reasoning step + tool call (CloudWatch GenAI dashboard).

Amazon Bedrock (Claude Haiku 4.5)

The reasoning model behind the agents, with extended thinking + structured output.

Amazon DynamoDB

The book (ocean-book: policies, claims, appetite, scenarios) + decisions (ocean-decisions, audit trail).

AWS App Runner + ECR

Hosts this Next.js app; ECR stores the container images.

MCP tool catalog

16 runnable · AgentCore Gatewayclick Run to call the real MCP tool — same path the agents use

Eligibility & KYC

gleif_entityVerify a cedent in the live GLEIF LEI registryGLEIF · live

sanctions_screenSanctions / adverse-media screenOpenSanctions / web · liveruntime-local

get_cedent_historyPrior decisions + relationship for a cedentMemory → DynamoDBruntime-local

claims_summaryHistorical claims/losses, total incurred + largestMCP → DynamoDB book

Hazard (live signal)

usgs_seismicEarthquake history near a place (M4.5+)USGS · liveruntime-local

web_searchWindstorm/flood outlook, market conditions, newsWeb · liveruntime-local

cyber_threat_signalInsured's stack vs actively-exploited CVEs + exploit probabilityMCP → CISA KEV + EPSS · live

Pricing, accumulation & capital

price_treatyTechnical price for a cat-XL layer (exposure + experience rating)MCP → pricing engine

query_accumulationMarginal 1-in-100 PML vs zone appetiteMCP → accumulation engine

capital_impactMarginal diversified Solvency II SCR + solvency ratioMCP → capital model

search_guidelinesSemantic search over underwriting guidelines (cited)MCP → Titan RAG

Cyber & portfolio analytics

cyber_accumulationAccumulation for treaties sharing a dependency vs its appetiteMCP → dependency engine

scenario_exposureSystemic scenario (Cloud Down / ransomware / mass-vuln) aggregateMCP → scenario engine

cyber_concentrationALL cyber dependency concentrations in one callMCP → concentration engine

exposure_summaryConcentration across all zones vs appetiteMCP → portfolio engine

capital_positionCurrent own funds, SCR, solvency ratio, headroomMCP → capital model

Data lookups

rds_scenariosRealistic Disaster Scenarios (Lloyd's RDS) net loss vs appetiteMCP → RDS engine

appetite_limitsAll appetite limits + utilisation (zones + cyber dependencies)MCP → DynamoDB book

treaty_lookupFull detail of one in-force treatyMCP → DynamoDB book

cedent_portfolioAll treaties + claims for one cedentMCP → DynamoDB book

Deterministic methods (the numbers)

Every figure in a recommendation comes from these engines — the LLM owns only the verdict, rationale and conditions, and each claim is cited to the tool that produced it.

Cat loss model

Lognormal severity → AAL + PML at 1-in-100 / 1-in-250 return periods.

Technical pricing

Blend of exposure rating + experience/burning-cost, risk load, cost of capital → ROL, payback.

Accumulation

Marginal PML contribution vs appetite — by CRESTA zone (nat-cat) or shared dependency (cyber).

Marginal SCR

Stand-alone capital diversified against the portfolio by correlation group → solvency ratio.

Cyber threat signal

Cross-references the insured's dependencies against CISA KEV + EPSS → live hazard view.

Systemic scenarios

Cloud Down / ransomware / mass-vuln footprints → aggregate modelled loss across the book.

What's real vs. simulated

Real: the full AWS stack above, the agentic orchestration, the deterministic engines, and the live external feeds (GLEIF, USGS, CISA KEV, EPSS, web).

Simulated: only the book itself (policies, claims, appetite) — a representative synthetic dataset in DynamoDB. OAK replaces it with their real book in the same schema, and everything above runs unchanged.