The practical steps insurers are taking to build the capabilities required for agentic AI — modern architecture, trusted data, governance frameworks, and disciplined experimentation that positions them for scale. How do you move from isolated pilots to something operationally meaningful?