Pie CEO says insurers need in-house AI tools to compete

AI is making software development a bigger competitive factor in commercial insurance, according to Pie Insurance co-founder and CEO John Swigart. He argues that carriers relying only on licensed vendor tools will struggle to stand apart, while AI-fluent developers can help insurers build proprietary systems without the large engineering teams once required.

Swigart sees bespoke underwriting models as valuable where carriers have genuine specialization, but warns that they require substantial scale, capital and data, including external data licensed from third parties. Small commercial insurance remains harder to model than personal auto because business operations, exposures and classifications vary widely. Next Insurance, by comparison, reported more than 600,000 policyholders and $548 million in 2024 revenue after roughly a decade of building underwriting and loss data.

Near-term productivity gains are expected in analytical, operational and agentic use cases, with Swigart urging leaders to start experimenting rather than over-plan deployments. Research cited by Deloitte found agentic AI deployments in insurance produced underwriting efficiency gains as high as 36% and claims cycle-time reductions near 40% in early implementations.

Firmographic data is emerging as a key opportunity, especially for workers’ compensation, where accurate business classification directly affects pricing. The National Council on Compensation Insurance maintains roughly 700 distinct classification codes nationally, and Swigart says AI-enabled data streams could help carriers detect risk signals and classification errors before renewal or claims events.