Crawford and Company, the Atlanta, Georgia headquartered claims management and outsourcing provider, routes every new artificial intelligence tool through a formal review before it reaches a live claim. Adjusters and claims specialists hold the approval decision, according to David Wright, chief AI, data and cloud officer at the firm.
Wright told Insurance Business that his team tests AI assisted workflows with cross functional groups of adjusters ahead of any wider rollout. Reviewers see a prototype of a modified interaction and assess whether the AI version helps them do the work. They have the option to reject it. Accuracy carries the heaviest weight in that review, and the team runs repeated tests to confirm output stays stable across multiple uses. When adjusters flag problems, the tool is pulled and refined until they approve it.
Wright pointed to reserve setting on longer running claims as an area where the validation bar would sit especially high, since adjusters estimate the eventual cost of a claim and set reserves against that estimate. Regulation shapes the caution as much as internal preference. Claims adjusting depends on licensing, expertise and individual judgment, and Wright said carrier and corporate clients want AI to augment adjusters, speed their work and free time for the hardest parts of a claim.
On the build versus buy question, Crawford favors developing internal capability while sourcing large language models from outside providers. Wright said engaging with the changes directly helps the company understand the logic behind them. Employees can already create their own agents using Microsoft 365 Copilot. When a worker builds something useful, Wright's team reviews security and privacy, then promotes the agent so other teams can use it. The stated goal is to turn hundreds of frontline workers into sources of AI ideas.
Source: Insurance Business - https://www.insurancebusinessmag.com/us/news/technology/crawfords-ai-chief-explains-claims-innovation-strategy-588347.aspx