FAQ - Business Insurance
Frequently asked questions about building AI/RAG/agent applications in the B2B insurance domain. Oriented toward developers constructing on this sector, not insurance consumers.
---
What is "business insurance" and how does it differ from retail insurance?
Business insurance (or commercial/industrial insurance) covers the risks of companies, organisations, and professionals. It differs from retail in the complexity of coverages (bespoke policies, high limits), the role of the broker as a technical intermediary, the negotiability of contractual conditions, and the underwriting process based on specific corporate risk data.
What are the main insurance lines relevant to SMEs?
Key lines include: General Liability, Product Liability, Professional Liability, Fire and Industrial Risks, Theft, Business Interruption (BI), Fleet Insurance, Employee Accident and Health, Trade Credit Insurance, and Surety Bonds. Manufacturing SMEs often add CAR (Contractor's All Risks) and machinery breakdown coverage.
How does the underwriting process work for a business policy?
The policyholder submits a risk questionnaire; the broker supplements it with financial data (turnover, headcount), technical data (type of operations, installations), and past claims history. The insurer's underwriter assesses the risk, calculates the premium using actuarial models, and issues an offer with specific terms. For complex risks, an on-site technical inspection is conducted.
What makes a business insurance claim "complex"?
Complexity depends on the claim amount (> €100,000), the number of coverages involved (e.g., fire + BI + third-party liability), co-insurance arrangements (multiple insurers), latency between event and loss manifestation (professional liability, product liability), and the need for specialist expert assessment (engineers, chemists, medical experts).
What are co-insurance arrangements and when do they apply?
Co-insurance involves multiple insurers sharing coverage of a risk, each for a fixed percentage share. Applied to large industrial risks where no single insurer is willing or able to assume the full exposure (e.g., petrochemical plants, large infrastructure projects). The lead insurer (delegatario) manages the policy on behalf of the others.
How is a Professional Liability premium calculated?
The premium depends on the professional's turnover or fees, the type of professional activity, the requested limit, the claims history over the past 3-5 years, and the geographical operating area. For some professions (physicians, lawyers), industry rating tables exist. The London market (Lloyd's) is often involved for high-exposure risks.
What does a Cyber Risk policy for a business typically include?
A cyber policy covers: first-party losses (system restoration, data recovery, crisis management, business loss), third-party liability (GDPR data breach liability, client service interruption losses), extortion (ransomware), legal expenses, and mandatory regulatory notifications (Data Protection Authority, CSIRT). Some policies include an incident response (IR) team.
How does Solvency II affect insurance company IT systems and APIs?
Solvency II mandates standardised reporting (QRT - Quantitative Reporting Templates) to IVASS/EIOPA in XBRL format. Insurers must maintain data governance systems and audit trails for SCR/MCR/ORSA calculations. Internal APIs manage data flows between actuarial, accounting, and risk management systems aligned with regulatory templates.
What is the difference between premium reserves and claims reserves?
Premium reserves (unearned premium reserves) represent the share of collected premiums relating to risks not yet matured at the balance sheet date. Claims reserves cover obligations for claims that have already occurred: partially paid, in the process of settlement (RBNS - Reserved But Not Settled), or not yet reported (IBNR).
How does reinsurance work and why is it relevant for AI systems?
Reinsurance transfers part of the risk and premium to another insurer (reinsurer). It can be proportional (quota share: a fixed share of every policy is ceded) or non-proportional (excess of loss: the reinsurer intervenes only above a retention threshold). AI pricing and risk management systems must model the impact of reinsurance on the net retention.
What is IBNR and how is it estimated?
IBNR (Incurred But Not Reported) is the value of reserves for claims that have already occurred but have not yet been communicated to the insurer. Estimated using actuarial methods: Chain-Ladder projects paid claims on development triangles; Bornhuetter-Ferguson combines the projection with an a priori loss ratio estimate. Machine learning models are increasingly being integrated with these classical methods.
What data is needed for an AI pricing model in fleet motor insurance?
Key factors include: vehicle type (category, age, power), usage (annual mileage, use type), operating location, driver profile (age, experience, claims history), and active/passive safety features. Telematics data (UBI - Usage-Based Insurance) enables real-time behavioural pricing.
How does an insurance system handle real-time reserve development?
Reserve development monitoring tracks the evolution of claims reserves against initial estimates. A real-time system receives updates from the claims management system, calculates the delta versus booked reserves, and updates predictive models. Actuarial dashboards display development patterns by line of business and accident year cohort.
What are parametric insurance products and how do they integrate with APIs/IoT?
Parametric policies pay a fixed amount upon exceeding a threshold measured on an objective index (seismic magnitude, rainfall millimetres, temperature). Integration with weather APIs (e.g., ECMWF, ClimaCell) or IoT sensors enables automatic settlement without expert assessment, reducing claims management costs and processing time.
What disclosure obligations apply to digital insurance product distribution?
The IDD (Directive 2016/97) requires: delivery of the IPID (Insurance Product Information Document) before subscription, suitability assessment (advice) or appropriateness check (execution-only), disclosure of conflicts of interest, remuneration, and complaints handling. Digital platforms must track document delivery and informed consent.
How is a credit risk scoring system implemented for trade credit insurance?
The system integrates: debtor balance sheet data (leverage, EBITDA coverage), behavioural information (historical delays, payment patterns), sector data (default rates by industry and geography), and agency ratings (S&P, Moody's, or SME-focused providers). The model predicts 12-month probability of default (PD), which feeds into pricing and the insurance credit limit decision.
---
Note: This document is descriptive and technical in nature for software developers. It does not constitute insurance or legal advice. For specific obligations, consult a registered insurance broker or qualified legal counsel.