AI agent training for underwriters in insurance in the Netherlands

Dutch insurance underwriters need training that prepares them to oversee agents on term life cover linked to mortgages, income protection for the self-employed, and property. The rules are precise: Dutch law limits the medical questions insurers may ask below certain sums, and the EU AI Act treats life and health pricing as high-risk. Underwriters must check what the agent used and decide every medical case.

What does an insurance underwriter in the Netherlands handle?

Term life cover is commonly taken out alongside a mortgage, so life underwriters see a steady flow of applications from intermediaries working to a house purchase deadline. Dutch law restricts the medical questions insurers may ask for cover below certain amounts and governs how medical examinations are used, so underwriters must apply those limits before they consider health at all.

The country's large population of self-employed workers creates demand for income protection cover, which is underwritten on occupation, income history and health, and is notoriously hard to price well. Property and liability underwriters handle household and small business risks, with flood and subsidence questions in some regions. Insurers also consult shared industry incident registers when assessing proposals, under the sector's own code of conduct.

Where do agents help, and how far may they act?

1. Medical question eligibility, depth 1 to 2. The agent checks whether the requested sum falls within the band where medical questions are restricted, and drafts the path: acceptance without health questions, or a health declaration for review.
2. Health declaration triage, depth 2. Where questions are permitted, the agent maps declared conditions to the insurer's guidelines and drafts the medical adviser's file. Decisions stay with people.
3. Self-employed income assessment, depth 2. The agent reads annual accounts and tax return data, calculates the insurable income under the insurer's rules and flags volatility. Accounts prepared by the applicant's accountant are signed; figures typed into the application are declared.
4. Property proposal enrichment, depth 2 to 3. The agent adds the insurer's own hazard data and rates standard household risks; a certified agent may accept standard ones within the insurer's limits.
5. Acceptance with exclusions or surcharges, depth 4. Held for an underwriter every time.

What must underwriters learn, and what stays with them?

Underwriters need to review the agent's work with the restrictions in mind: how underwriters review an agent's decisions sets out the routine. For a Dutch life file, the first check is whether the agent asked anything it should not have, given the sum insured. The record shows which questions were presented and which data the agent read.

The second skill is improving the agent. Income protection is where agents make plausible errors, such as treating a one-off contract payment as recurring income. When an underwriter corrects it, the reason should state the rule so the agent's sources change. Teaching underwriters to bank corrections gives the method.

Underwriters keep:

• every decision on health grounds, with medical adviser input;
• insurable income judgements for volatile self-employed earnings;
• use of shared incident register findings;
• acceptance outside guidelines.

The firm limits are in what underwriters should never let an AI agent do, and why underwriters are the real control explains why that matters.

Which Dutch and EU rules shape oversight and literacy?

De Nederlandsche Bank supervises insurers prudentially and the AFM supervises conduct, including whether products serve customers' interests and are sold with care. EIOPA promotes convergence across the EU. Both Dutch supervisors will expect an insurer to show who oversees agents that draft or make underwriting decisions and how those people are qualified.

The EU AI Act applies directly. Annex III lists risk assessment and pricing for natural persons in life and health insurance as high-risk, which covers term life and income protection underwriting. Deployers must meet Article 26 obligations, give human oversight under Article 14 to competent people who can override, and in some cases carry out a fundamental rights impact assessment under Article 27. Article 4 requires AI literacy among staff.

GDPR governs proposal data, with health data as special-category data, enforced by the national data protection authority. Article 22 restricts decisions based solely on automated processing with significant effects, and Article 35 impact assessments are expected for this kind of processing. Dutch law on medical examinations for insurance, and the sector's code of conduct on personal data, add specific limits that must be loaded into the agent's sources. DORA applies to insurers from 17 January 2025.

This is an explanation of the rules, not legal advice; confirm how they apply to you with counsel.

A 90-day plan

Weeks 1 to 4. Foundations for life, income protection and property underwriters and for medical advisers' support staff. Pryme Intelligence Academy's Foundations track is free and tested in insurance, and starts the Article 4 record. Inventory agent capabilities and mark the high-risk ones.

Weeks 5 to 8. Replay a quarter of term life applications at depth 1 and check every file for medical question limits. Replay income protection files and compare insurable income figures. Correct the agent's sources. Approvers sit the operator exam.

Weeks 9 to 12. The agent drafts at depth 2 on live files. Two-week practicum, then calibration on twenty income protection files. Approval rights for held acceptances pass to certified underwriters. The full role map is in AI agent skills by role.