AI in insurance. For insurers and insurance brokers.
Claims intake, underwriting support, policy documents and customer service with Claude, with decisions kept with people, documented for DORA and governed under the EU AI Act.
30 minutes with a solution architect.
What is AI in insurance, and how does Solorai help?
AI in insurance is the use of AI models such as Claude to take in claims, support underwriters, check policy documents and answer customers, with people responsible for every claims and underwriting decision. Solorai sets up Claude Enterprise for insurers and brokers, builds AI agents for insurance workflows and governs their use under the EU AI Act, DORA and the expectations of supervisors.
Written and reviewed by Solorai
- to assess your use cases and data
- 2 weeks
- people, at most, in a first pilot
- 50
- to roll out to up to 250 people
- 6 to 8 weeks
What can insurers and brokers do with Claude?
The six uses insurers usually start with. Claims handlers and underwriters stay responsible for every decision that affects a customer.
Insurers
Claims intake
Read claim forms, invoices and expert reports, check them against the policy and prepare the file for a claims handler.
Commercial lines
Underwriting support
Summarise submissions, extract the risk data and flag what is missing before the underwriter decides.
Insurers and brokers
Policy documents
Compare policy wordings, find the differences between versions and answer questions about cover with the clause cited.
Insurers and brokers
Customer service
Draft answers to policyholder questions about cover and claim status, in the customer’s language, for an agent to review.
Claims teams
Checks for inconsistent claims
Flag claims with inconsistencies for the investigations team, with the reasons stated. People decide what happens next.
Insurers
Regulatory reporting support
Collect the data and draft narrative sections of supervisory reports for the actuarial and compliance teams to check.
Confidentiality and compliance
How does AI in insurance stay fair and compliant?
Supervisors expect fair treatment of policyholders and sound governance for every AI system. These are the controls we put in place before the first claim goes into Claude.
Data under your control
Work runs in your own Claude Enterprise organisation, with single sign-on, roles and data policies. Under Anthropic’s commercial terms, your content is not used to train models by default.
People decide
Claims, underwriting and pricing decisions stay with people. AI agents prepare the file and ask for approval before anything reaches a customer.
Documented for DORA and your supervisor
We document the Claude service for your DORA register of information and map each use to the AI governance principles EIOPA set out in August 2025.
Classified under the EU AI Act
We classify each use before it goes live. Risk assessment and pricing for individuals in life and health insurance is high-risk.
Unacceptable risk
Banned since 2 February 2025, such as social scoring or emotion recognition at work.
High risk
- Risk assessment and pricing for individuals in life and health insuranceHigh-risk obligations, Annex III, point 5(c)
- Screening job applicantsHigh-risk obligations, Annex III, employment
Limited risk
- A chatbot that answers policyholdersTell users they are talking to an AI system
Minimal risk
- Claims intake, policy checks and drafting for staffAI policy and AI literacy measures for staff
Point 5(c) covers life and health insurance only. Obligations for the high-risk uses listed in Annex III apply from 2 December 2027, as amended by Regulation (EU) 2026/1744. We classify every use before it goes live.
How we deliver
How does Solorai deliver AI in insurance?
The same three services as in every engagement, set up for the rules of this sector.
Claude Enterprise
Claude for claims, underwriting and service teams: single sign-on, roles, data policies, connectors to your core systems and training on real files.
Claude Enterprise servicesAI agents
One agent per workflow, such as first notice of loss intake, submission triage or policy wording comparison, monitored, logged and approved by a person.
AI agent servicesEU AI Act and governance
An inventory of every AI use, risk classification under the EU AI Act, an AI policy aligned with EIOPA’s principles and training for staff.
EU AI Act compliance services
Where to start
AI readiness assessment
About 2 weeks
We map your lines of business, documents and core systems, and pick the uses with the clearest value.
Pilot
Up to 50 people
One claims or underwriting team works on real files, with success criteria agreed before the start.
Rollout
6 to 8 weeks
Claude goes to the teams in scope, with connectors, playbooks and training.
FAQ
AI in insurance, in short
What is AI in insurance?
AI in insurance is the use of AI models such as Claude to take in claims, support underwriters, compare policy wordings, answer customers and prepare reports. Claims, underwriting and pricing decisions stay with people.
How do insurance companies use AI?
Insurers usually start with claims intake, submission triage for underwriters, policy wording comparison, customer service drafts and checks for inconsistent claims. Each use is classified under the EU AI Act before it goes live.
Is AI in insurance high-risk under the EU AI Act?
Risk assessment and pricing for individuals in life and health insurance is high-risk under Annex III, point 5(c), with obligations from 2 December 2027. Claims intake, policy checks and drafting are not high-risk.
What does EIOPA expect from insurers using AI?
In its Opinion of 6 August 2025, EIOPA explained how existing insurance rules, such as Solvency II, the Insurance Distribution Directive and DORA, apply to AI systems: a risk-based and proportionate approach covering areas such as fairness, data governance, transparency and human oversight.
Does DORA apply to insurers?
Yes. Insurance and reinsurance undertakings are financial entities under DORA, which applies since 17 January 2025, so AI services such as Claude go into the register of information like any other ICT service.
How does an insurer start with AI?
With an AI readiness assessment of about 2 weeks, then a pilot with one claims or underwriting team of up to 50 people, then a rollout of 6 to 8 weeks to the teams in scope.
Related reading
- Service
Custom AI agent development
Custom AI agent development: requirements, architecture, tool integrations, evaluation and monitoring, built for your data, systems and security rules.
See the service - Service
AI literacy training under Article 4 of the EU AI Act
AI literacy training for staff under Article 4 of the EU AI Act: role-based programmes, practical exercises and records you can keep for compliance.
See the service - Guide
EU AI Act checklist for companies
A step-by-step EU AI Act checklist for companies: inventory your AI systems, classify risk, map obligations by role and set deadlines.
Open the checklist
Sources: Regulation (EU) 2024/1689, Regulation (EU) 2026/1744
AI for your insurance business. Start with a 30-minute briefing.
Tell us about your lines of business and your goals. We will reply with a clear next step.