AI in insurance is more than just a tool. It changes how work gets done in underwriting, claims, special investigations, customer service, distribution support, actuarial analysis, compliance, IT, and HR. Success comes from people using AI safely, consistently, and measurably—not from having the most advanced models.
McKinsey points out that how well AI is adopted and managed often determines whether it remains unused or truly changes operations. They suggest setting aside a budget for adoption that matches the cost of developing the technology.
How do you train your workforce to work with AI without just offering a one-time “GenAI 101” webinar that people soon forget?
Most Roles Stay with AI
NAIC materials show the integration of AI into claims handling and chatbots, making it part of daily work for human employees. Training should include:
- The Decision-making process, including where and why humans confirm and verify AI decisions
- How to spot hallucinations, bias, drift, and leakage
- How to document and escalate issues
Build an AI Training Program with the Same Rigor as a Compliance Program
Regulators and supervisors are making it clear that AI governance must have structure, accountability, and controls. The NAIC’s Model Bulletin requires insurers to have a written AI governance program that meets risk management, accountability, and compliance standards.
In the EU, EIOPA has also provided guidance on AI governance and risk management in insurance. While governance sets the rules, workforce training ensures people follow them safely.
A practical training approach aligns with recognized risk frameworks like the NIST AI Risk Management Framework (AI RMF 1.0), which focuses on trustworthiness traits such as validity, safety, security, accountability, transparency, explainability, privacy, and fairness through bias management.
A 3-layer training model that works well in insurance
1) AI Literacy for Everybody
Create basic safe behaviors that include:
- What AI can and can’t do in claims, underwriting, and service
- IT rules for PHI, PII, confidential business info, and IP
- Not acceptable AI output with adverse actions, such as coverage decisions, denial rationales, and regulatory communications
- Examples of common hallucinations, overconfidence, and subtle bias
- Issues and near-misses reporting
This start is critical because skill disruption is real, and we expect ongoing learning across industries. The World Economic Forum’s Future of Jobs report highlights the constant changes in skills and the need for continuous upskilling and reskilling.
2) Most Staff Receive Role-based ‘AI Workflow’ Training
Show employees working with AI how it positively changes their workflows by starting new learning paths by job function:
- Claims and operations – intake summarization, extracting documents, next-best-actions, flagging fraud, drafting customer messages, and Quality Assurance
- Underwriting – policy submission triage, appetite matching, guideline lookup, and tracing decision support
- Actuarial and analytics – code assistance, scenario narration, model monitoring workflows, data quality checks
- Legal/compliance/risk – AI use approvals, vendor risk, model documentation, audit trails
- IT/security – secure deployment patterns, access controls, monitoring, incident response
3) AI Specialists Do In-Depth Training
Design AI capabilities that reduce dependence on outside vendors and consultants in these functions:
- AI insurance domain and process redesign owners
- Model risk and validation professionals
- Prompt and automation builders who focus on secure workflows
- Data stewards for AI quality, lineage, and permissions
- AI governance policy, approvals, and controls leads
See ‘Human-in-the-Loop’ as a Trained Skill
The biggest risks in insurance come from AI’s influence on decisions that affect customers, such as eligibility, pricing, claim outcomes, and investigations. Training should cover:
- Verification behaviors – how to verify against source documents
- Escalation thresholds – what requires a supervisor and compliance review
- Documentation habits – capturing when a human agrees and disagrees with AI
- Bias awareness – understanding protected classes, proxy variables, and disparate impact risk
Governance expectations connect directly to training. NAIC’s model bulletin looks at accountability, compliance, transparency, and fairness as the core principles that guide daily workforce actions.
Change Management Focus on Training Drives Adoption, Not Just Attendance
Insurance companies train everyone, and see mixed results because workflows, incentives, and risk acceptance don’t change. Use proven strategies from successful transformations:
- Set money aside for adoption activities such as communications, redesigning workflows, and change champions, not just for technology.
- Create departmental AI leaders because peer support works best in change implementation
- Roll out AI in stages in a sandbox, a supervised pilot, then a full-scale program.
- Track usage and results, such as cycle time, fewer errors, customer satisfaction, rework rates, and mistakes.
A simple 90-day rollout plan for insurers
Days 1 to 15: set the rules and establish the baseline
- Publish AI use policy and information management do/don’t
- Deliver 60–90 minute AI literacy training to all staff
- Stand up a single “AI help desk” channel and escalation path
- Define “no-AI zones,” like certain regulated communications, and “AI-assisted zones.”
Days 16 to 45 – pick three workflows and provide role-based training
Choose workflows with templates like claims note summarization, inbound document triage, and customer email drafting, with:
- Accessible and permissioned source data
- Clear success metrics
- Naturally present human review
Days 46 to 90 – scale governance and measurement
- Launch department champions program
- Add logging, approvals, and QA sampling checkpoints
- Create “approved prompts,” “approved workflows,” and “when to escalate.”
- Begin product owner, model risk, and data stewardship
Define ‘Good’
You know training works when you see:
- Higher throughput without rising error or rework
- Documented clear audit trails
- Fewer unsafe employee use of random tools mimicking AI behaviors
- A shared vocabulary across operations, IT, risk, and compliance
- Managers coach AI use like any other performance metric
Welcome to the future of insurance that runs at the speed of now. Agility Holdings Group (AHG) invests in innovative InsurTech, HealthTech, and related companies that aim to revolutionize access to insurance products, establish patient care, and improve health outcomes.
Please visit our LinkedIn page for more information about AHG.