Insurance companies are putting artificial intelligence to work in claims, fraud detection and other core operations, but most may be overestimating how far their AI programs have progressed, according to EXL’s 2026 U.S. Enterprise AI Study.
Seventy-six percent of insurers surveyed believe they are ahead of competitors in AI, yet only 6% met EXL’s criteria for an AI "Leader." Another 72% were classified as Followers, the largest share of companies in that middle category among the industries surveyed. Scaling AI is now a high priority for 96% of insurers, up from 86% in 2025.
Claims is already one of the more common applications. Forty-two percent of insurers reported using AI in claims, behind fraud detection and customer servicing, both at 54%, financial crime compliance at 44% and risk management at 44%. Insurers are also beginning to redesign broader workflows around the technology, including processes that connect intake, conversational AI and claims handling at first notice of loss.
The findings suggest claims organizations are moving beyond experiments with individual AI tools toward integrating the technology into the claims process itself. That shift has practical implications for adjusters because the success of AI-assisted claims handling depends not only on the technology, but also on the quality of the underlying data, how workflows are designed and which decisions remain subject to human review.
Insurers appear relatively successful at getting AI projects out of the pilot stage. Sixty-two percent of insurance AI pilots reach production, the highest rate among the industries surveyed. EXL found successful projects tend to have a clear business owner, a defined outcome and agreement on what qualifies as production-ready.
Agentic AI, which can perform and coordinate tasks across systems with greater autonomy than conventional AI applications, is also gaining ground. Risk management leads insurance agentic AI use at 54%, followed by actuarial, underwriting and pricing at 46% and customer experience at 45%. The broader EXL report says four in 10 companies across the surveyed industries have moved agentic AI beyond the pilot stage.
The technology is producing measurable returns where insurers have deployed it. Nearly half of insurers, 46%, have fully deployed AI in actuarial and underwriting. EXL found AI Leaders generated 40% more revenue growth and 37% more cost reduction than Laggards in use cases where AI was applied. The broader report also shows insurers reporting sizable cost, revenue and margin improvements from both conventional and agentic AI applications.
Data remains a significant obstacle. Nearly 92% of insurers said their data presents a challenge to AI success, and insurers identified data silos as a barrier more frequently than companies in any other industry surveyed. Only 24% consider themselves leading-edge in data management maturity, while 38% completely agreed that their organizations have sufficient governance for ethical and responsible AI use.
Those weaknesses are particularly relevant to claims operations. Claims systems draw on policy information, loss reports, photographs, repair estimates, medical records, third-party data and adjuster documentation. An AI system’s usefulness can be limited when that information is incomplete, inconsistent or trapped in separate systems. Governance also becomes more important as AI moves from helping adjusters retrieve and summarize information toward influencing decisions affecting claim outcomes.
The report’s broader findings support a model in which AI handles more routine processing while humans concentrate on exceptions and decisions requiring judgment. A case study on page 8, although involving finance rather than insurance claims, describes an agentic AI system that matches documents, assigns confidence scores and sends exceptions or low-confidence results to employees for review. The example illustrates one potential approach for claims organizations seeking automation without removing human oversight from uncertain cases.
For adjusters and claims managers, the study points to a transition that is already underway rather than a distant replacement scenario. AI is entering claims workflows, but the relatively small number of insurers EXL considers AI Leaders indicates that deploying tools is not the same as achieving mature, enterprise-scale AI. Data quality, governance, workflow design and decisions about where human judgment is required remain central to how far claims automation can advance.
