AI in Insurance

The insurance sector is undergoing one of the most profound transformations in its history. Artificial intelligence (AI) is no longer a theoretical innovation or a topic reserved for futurists, it has become a concrete, operational capability that is rapidly reshaping how insurers work, make decisions, and create value.

During our session hosted at the KPMG in Belgium offices on 24 November, we had the pleasure of welcoming members of ACAM–VMVM for a deep dive into the real impact of AI on insurance operations, distribution, governance, and risk management.

Across presentations, live demonstrations, and discussions with industry leaders, six key takeaways clearly emerged.

One of the strongest messages of the morning was simple: AI is not a future ambition, it is already widely used across the insurance value chain. From claims triage and fraud analytics to underwriting support, document processing, customer engagement, and contact center optimization, AI tools are now part of daily operations in most insurers.

Concrete use cases presented during the session illustrated this reality:

• Automated classification of emails and documents using natural language processing.
• FNOL acceleration through automated segmentation, extraction of critical data, and accuracy rates above 90%.
• Real-time customer assistance through AI-augmented call centers, reducing call duration by up to 30%.
• Enhanced fraud detection models able to detect anomalies and suspicious patterns far earlier.

The variety of tools from generative AI to predictive models to image-based algorithms shows that AI is not one technology but a wide spectrum of capabilities supporting operational efficiency and customer experience.

While the appetite for AI is strong, the level of maturity across organizations varies significantly.

Survey data shared during the event revealed that:

• Many insurers still face low internal AI literacy levels.
• Concerns around premature investment and technology obsolescence persist.
• Governance and regulatory clarity remain top-of-mind questions.

However, the strategic opportunity is unanimously recognized, with 85% of insurers believing that companies embracing AI early will gain a decisive competitive edge. The challenge is no longer why but how to prioritize, how to de-risk, how to scale, and how to align investments with business value.

A central part of the session focused on the next generation of AI: AI agents. These systems go far beyond traditional automation:

• They analyze context;
• They plan sequences of actions;
• They coordinate tasks;
• They interact across systems; and
• They learn and optimize over time.

We explored the TACO Framework (Taskers, Automators, Collaborators, Orchestrators), which classifies AI agents based on their complexity and level of autonomy.

Practical insurance applications are emerging at high speed:

• Intake agents capable of analyzing claims declarations, extracting data, checking coverage, and drafting communications.
• Coverage and liability agents autonomously coordinate multiple models to reach a preliminary assessment.
• Fraud agents assisting handlers by combining structured and unstructured data in seconds.
• Operational orchestrators that could eventually manage entire end-to-end workflows.
• This shift opens the door to a blended workforce where people and AI work side by side, with “digital colleagues” taking on routine, data-heavy, or multi-step tasks.

With innovation comes responsibility. The upcoming European AI Act introduces new obligations for insurers as both deployers and providers of AI systems. High-risk use cases such as creditworthiness assessments, pricing models, or claims decisioning will fall under strict requirements:

• Documentation
• Data governance
• Monitoring
• Human oversight
• Robustness and accuracy
• Bias and discrimination safeguards
• Transparency obligations

Insurers will need stronger governance frameworks, clearer accountability, and consistent oversight mechanisms. This includes establishing lightweight controls for low-risk use cases and more advanced processes for high-risk systems.

The message was clear: trusted AI is not optional; it is a regulatory and ethical foundation for sustainable AI adoption.

Almost all insurers experiment with AI, but few have fully scaled it across the organization. The session highlighted a structured three-phase maturity journey:

Building foundations: literacy, strategy, cloud pilots, initial models.

Integrating AI into core workflows: claims, underwriting, call centers, customer support.

Reinventing business models and ecosystems including dynamic pricing, real-time risk management, personalized prevention programs, and advanced agent-driven operations.

Value accelerates when organizations move beyond proof-of-concepts and adopt enterprise-level thinking.

Technology is evolving quickly, but the decisive enabler remains leadership. Insurance executives must take ownership of:

• aligning AI with strategic priorities;
• investing in talent and skills;
• orchestrating governance;
• ensuring regulatory readiness; and
• fostering a culture where humans and AI reinforce each other.

AI is no longer an IT topic. It is a business transformation driver that requires active leadership across actuarial teams, claims, underwriting, operations, risk, compliance, and the C-suite.

The ACAM–KPMG session confirmed a simple truth: AI is transforming insurance faster than any previous wave of technology, and the winners will be those who approach it with structure, ambition, and trust.

Insurers who combine innovation with governance, and experimentation with strategy, will unlock the full value of AI while reinforcing customer trust and operational excellence.