Kathmandu. Only one in every 10 insurance companies in the world has partially or fully implemented an AI agent. While another 20 percent of insurers are running pilot projects.
This has been seen in the new survey report released by the Capgemini Research Institute. According to the report, 42 percent of insurers have not yet developed a formal strategy to deploy AI agents in their operations.
Capgemini has also highlighted the growing interest in agentic AI in the insurance sector. That’s part of a wider $450 billion opportunity.
However, the study shows that most insurance companies are still in the early stages of adopting AI and many lack a clear implementation roadmap.
In key areas of 53% sales and customer service and 57% support, insurers expect ai agents to manage at least one process or subprocess daily. Despite these expectations, the insurer’s trust in the AI agent still remains low.
The overall confidence level of using AI in the insurance sector was an average of 54 percent in 2024. However, it has come down to 47 per cent in 2025.
Privacy (50 per cent), bias (50 per cent) and transparency (51 per cent) are the biggest risks for insurers when it comes to AI agents. Looking to the future, 26 percent of insurers believe that an AI agent will replace human employees within one to three years. According to the report, 35 percent of insurers expect these agents to work autonomously within the human inspection team.
