Perpendo Agent Platform

Foundation for agent planning, tool execution, observability and continuous governance—abstracting complex cores while preserving evidence and explainability.

Composable primitives for insurance automation across P&C, Global Specialty, L&A, and Reinsurance with compliance‑in‑the‑loop.
Agent Platform
Composable, governed agent runtime with deep-core integrations.
SVG → Table (pixel-like)
Data & Tools • Core connectors • DMS/CRM & rating • ACORD/PDF/images • Optional ECDIS hooks
Reasoning & Planning • Task decomposition • RAG/tool-calling • PII minimization • Chain-of-authority
Guardrails & Observability • Prompt/action lineage • A/B harness • Issue mgmt • Remediation
Compliance-in-the-Loop • Fairness & explainability • ECDIS registry • Provenance & approvals • Adverse action templates
Material decisions remain human-accountable. Agents provide decision support and supervised automation with full observability and rollback safeguards.
Data & Tools
Reasoning & Planning
Guardrails & Observability
Compliance ‑in‑the‑Loop
Compliance (CITL)
Compliance‑in‑the‑Loop turns policy into controls and evidence.
Policy → Controls
• Map NAIC/NYDFS/Colorado
• Control catalogs • RACI
• Policy versioning • exceptions
Risk, Bias & Quality
• Pre/ongoing impact tests
• Dataset/feature provenance
• Outcome thresholds • alerts
Model & Agent Registry
• ECDIS & model inventory
• Attestations • model cards
• Lineage: prompts/tools/data
Evidence & Audit
• Decision narratives • I/O
• Explainability bundles
• Regulator-ready reports/APIs
Vendor Governance
• Security/privacy questionnaires
• Service KPIs • logging
• Third-party ECDIS approvals
Workflow & Issues
• HITL checkpoints
• Issue capture • remediation
• RBAC • SoD • audit trails
Compliance‑in‑the‑Loop (CITL) turns policy into controls, continuously producing regulator‑ready evidence.
Policy → Controls
Risk, Bias & Quality
Model & Agent Registry
Evidence & Audit
Vendor Governance
Workflow & Issues

Line‑of‑Business Views

Compliance‑in‑the‑Loop turns policy into controls and evidence.

Core Platform Connectors

Governed automation with evidence.

Quote comparison explainability packs

Governed automation with evidence.

SIU signal surfacing with audit trails

Governed automation with evidence.

Bordereaux QA & treaty analytics

Governed automation with evidence.

Referral rationale with evidence citations

Governed automation with evidence.

Complex document pipelines (SOV, binders)

Governed automation with evidence.

Suitability scoring & adverse language

Governed automation with evidence.

Policy servicing agents (loans, 1035) with HITL

Governed automation with evidence.

Closed‑block conversions with lineage

Governed automation with evidence.

Cat/event roll‑ups & driver explanations

Governed automation with evidence.

Treaty program registry with ECDIS

Governed automation with evidence.

Exposure quality & exception analytics

Governed automation with evidence.
Outcomes & Assurances

Measured improvements with regulator‑ready artifacts.

Cycle time
32%
NIGO
28%
Leakage
19%
STP
24%
Audit completeness
100%
Material decisions remain human‑accountable. Agents provide decision support and supervised automation with full observability and rollback safeguards.

Ready to operationalize Agentic AI—safely?

Launch with connectors, govern with CITL, and ship audit‑ready evidence from day one.