· Property & Casualty Insurance · S&P MidCap 400
KNSL KINSALE CAPITAL GROUP INC $7.7B
No. Tools around the underwriter. Pricing with AI is still a goal.
6 of 15
Reach, Weight, and Proof.
Last updated score 30 SepLast review 29 OctLast earnings call 22 OctNext earnings call not posted yet
What they are doing
Kinsale, an excess-and-surplus specialist, says its pricing models stay proprietary and that underwriting judgment is the business. On the January 2026 investor day, current AI uses include software delivery, liability research, training, form development, and a claims-supervisor tool for high-loss claims. Segmenting and pricing risk with AI is listed as a goal, alongside future underwriting insights. On the 23 October 2026 call, Salmaan Allibhai said generalized linear models and other machine-learning algorithms feed the rates, and the rating engine produces a technical price for every risk. Underwriters use that price. He pointed at the book loss ratio as the result. That is the whole book, not a measured lift from the model. Coding assistants do not count. The published score is still 6. The recommendation, not yet accepted, is Reach 2, Weight 3, Proof 3, total 8.
Why the score
- Reach 2 not 3: software delivery plus claims and underwriting support. Not the pricing decision.
- Weight 2 not 3: the tools assist people. The company says judgment is still the business.
- Proof 2 not 3: named on the investor day. The pricing use is a goal, and no result is given.
What would move it next quarter: A loss figure that isolates the pricing model from the rest of the book. The technical price for every risk will not, by itself, move Proof to 4.
Open these first
Evidence
Kinsale says intentional use of AI has accelerated software delivery. Generative AI goals include improving how the company segments and prices risk. Listed current uses include liability research, training, and form development. A claims tool helps supervisors watch high-loss claims. Underwriting insights are described as a future use.