The Doubling Question

• Markel writes roughly $10 billion of the $700 billion global specialty insurance market. Wilson’s read: that’s room to grow, not a ceiling.
• His management approach: instead of asking a business leader to grow 5-6%, he asks how they’d double the business in five years — surfacing bigger ideas rather than incremental optimization.
• New categories are expanding the specialty pool itself: data center risk, satellite and space exposure, and marine war coverage in the Persian Gulf are recent, fast-growing additions.
• His guardrail on growth: “Top line is vanity, bottom line is sanity.” The doubling mindset is paired with underwriting discipline, not a trade against it.

Specialty Insurance’s Moment

• Wilson draws a parallel to media: just as audiences moved from four TV channels to infinite, personalized content, commercial insureds increasingly want coverage built around their specific risk.
• He’s candid that insurance carries a unique challenge as a product: people don’t want to buy it and don’t want to use it, which puts pressure on specialty insurers to earn the sale.

Rebuilding AI End to End, Not Task by Task

• Markel Insurance restructured into 14 business units (roughly 60 specialisms), each with a single accountable leader charged with reimagining product and process around AI.
• Wilson’s caution against partial AI adoption: applying it to only one step of a process without rethinking the whole thing creates a bottleneck, “like putting a rocket jet on the back of a bicycle.”
• Cortex, a new AI-native business built end-to-end on Claude with a pricing engine called Hyperexponential, was built in 12 weeks and targets difficult-to-place US casualty risk.
• Markel’s Warranties & Indemnities team in London used the legal AI model Harvey to cut the time to reach a decision-ready view on an M&A data room from about a month to roughly seven minutes, with accuracy Wilson estimates at 97%, a five-point improvement in underwriting pick, and premium in that line more than doubling over the past year.
• In one case, the AI flagged a newly filed class action lawsuit relevant to a $100 million financial institutions risk that the human team hadn’t caught.

Guardrails, Culture, and Reverse Mentoring

• Every large language model Markel staff use runs inside Markel’s own environment so company data doesn’t leave the firewall — inside that boundary, teams have wide latitude to experiment.
• The Accelerator Fund put $1 million behind ten employee-pitched AI ideas; Wilson expects more than $10 million in returns from that investment in its first year.
• Wilson credits reverse mentoring — biweekly sessions with a junior colleague — with reshaping how he thinks about AI and how the next generation approaches problem-solving.

AI as an Equalizer for Small, Specialized Businesses

• Wilson argues AI has democratized technology for niche specialty lines previously too small to justify major core-system investment, moving decisions closer to the underwriters who understand the customer.