AI in Insurance Operations: Failures, Wins, and a Playbook for New Use Cases

Artificial intelligence is everywhere in insurance conversations—but successful implementation depends less on hype and more on selecting the right problems to solve.

In IASA’s April 8 webinar, Oscar Martinez, Founder and CEO of Velos, shared a practical framework for evaluating AI automation opportunities in insurance operations. Drawing from real-world deployment successes and failures, Martinez explored why some AI initiatives deliver meaningful operational improvements while others become costly false starts.

Rather than focusing solely on technology, the session emphasized the importance of process design, documentation, and operational consistency when evaluating automation opportunities. Martinez encouraged organizations to think beyond the question of “Can AI do this?” and instead ask whether a process is stable, measurable, and clearly defined enough to support automation successfully.

Watch Key Takeaways

Recap of the IASA Webinar held April 8, 2026

Common Patterns Behind AI Successes—and Failures

Throughout the webinar, Martinez highlighted several recurring themes that influence whether AI implementations succeed in insurance operations. One major challenge is attempting to automate processes that lack clear workflows or standardized rules. In many failed implementations, teams discovered that the process itself was inconsistent long before the AI technology became the issue.

The session also explored the risks of trying to fully automate highly subjective work. In some cases, AI proved highly effective as a research or decision-support tool, while fully autonomous execution remained difficult because “good” outcomes still relied heavily on human judgment and experience.

At the same time, Martinez shared examples of successful automation initiatives involving high-volume, rules-based operational tasks such as invoicing, reconciliation, and reporting workflows. In these scenarios, organizations were able to improve processing speed, increase scalability, and reduce manual effort by combining clear standard operating procedures with continuous accuracy monitoring and refinement.

A Practical AI Evaluation Framework

A key takeaway from the session was the importance of evaluating automation opportunities through a structured lens. Martinez introduced a scorecard-based approach focused on several factors, including:

• Process stability and consistency
• Clearly defined outcomes and measurements
• Exception rates and edge cases
• Operational pain points and economic impact
• Ability to document workflows through standard operating procedures (SOPs)

The webinar emphasized that successful AI deployments are often less about adopting the newest technology and more about matching the right solution to the right operational challenge.