Executive summary

This article introduces Bernoly’s AI value extraction framework for insurance, arguing that AI creates measurable business value only when it improves real operational workflows. Rather than treating AI as a standalone tool, insurers should identify high-value workflows, decompose human work into cognitive tasks, and redesign processes around structured human–AI collaboration.

The paper focuses on two phases: Placement, which determines where AI can create value, and Intelligence Workflow Architecture, which designs how humans, rules, models, and AI agents work together to improve economic outcomes.

‍

What you’ll read

• The source of AI value: Why workflow transformation, not the AI model itself, is the true driver of measurable impact.
• Workflow Lake: How insurers can capture actual workflows at sufficient granularity to make AI redesign possible.
• Human Task Ontology: A method for breaking human work into cognitive components such as information retrieval, reasoning, decision-making, action, monitoring, and learning.
• Intelligence Workflow Architecture: How insurers can redesign workflows by combining humans, rules, predictive models, AI agents, and multi-agent systems.
• Human–AI collaboration models: A practical role-based model defining who triggers work, who decides, and who acts across human and AI agents.
• AI learning optimizers: Why AI-enabled workflows can appreciate in value over time when feedback loops and outcome data are captured effectively.

‍

"Insurers will not win with better AI tools alone; they will win by rebuilding the workflows where AI, humans, and decisions meet.”

‍

‍