The Columbia Insurance Kalepa agreement is a five-year commitment announced on 17 November 2026, under which Columbia Insurance, a property and casualty carrier for small and mid-sized businesses in 14 states, will run underwriting work on Kalepa’s AI platform. Fintechbits checked Kalepa’s newsroom and found that Columbia is the third carrier selection it lists for 2026, after James River on 10 February 2026 and Church Mutual’s CM Vantage unit on 3 September 2026, and the only one of the three whose announcement states a contract term. Neither company disclosed a price, a premium figure, a start date or the lines of business involved.
What the Columbia Insurance Kalepa agreement covers
Kalepa’s announcement says the platform will take over activities such as submission intake and clearance, and support triage, risk selection, rating and portfolio management. Work that AI can do accurately and consistently is automated, and underwriters keep complex risks and agent relationships. The Columbia Insurance Kalepa release gives no volume target, no list of lines, and no schedule.
It does name a goal. Columbia wants stronger underwriting decisions, less back-office effort and what it calls profitable premium growth. The company says it will work with consulting and systems integration partners, but does not name them. A five-year term for a carrier that sells through independent agents is a long commitment to a software category that is changing fast.
Columbia’s own website lists an A.M. Best rating of A/Stable and serves seven small-business sectors, from contractors and restaurants to wholesalers. It did not mention Kalepa on 17 November 2026.
How Columbia made the choice
The Columbia Insurance Kalepa selection followed what Columbia calls an extensive evaluation that brought technology, operations and underwriting leaders together. Chief underwriting officer Keith Maciejewski said the team stopped treating it as a systems checklist and asked “which platform will actually help us drive gross written premium growth and improve our loss ratio?”
That framing is the most useful detail in the announcement, because it sets the measure Columbia will judge Kalepa by. Premium growth and loss ratio are results that show up in statutory filings and rating agency reviews, not in a demo. It also means Kalepa has accepted an outcome-based test, even if no figures were attached.
Governance carried weight too. Chief information officer Lisa Wharton said that as a regulated insurer Columbia applies “a high bar for governance, oversight, traceability and data protection”. For a carrier that answers to state insurance departments for underwriting decisions, traceability of each AI-assisted decision is a requirement, not a feature.
Where the Columbia Insurance Kalepa deal sits among 2026 customers
Next to the Columbia Insurance Kalepa announcement, the other two 2026 selections on Kalepa’s newsroom are both excess and surplus businesses. The James River announcement covers E&S underwriting, and Church Mutual chose the platform for CM Vantage, its E&S carrier. Columbia’s description is comprehensive property and casualty cover for small and mid-sized businesses, with no E&S focus stated.
Kalepa has momentum behind it. Fintechbits reported that the company raised $75 million at a $470 million valuation on 5 November 2026, and that it plans to double its team in 2027. Investors are backing AI underwriting from other angles too: Great Hill Partners took a stake in Aurenity, an E&S managing general agent, as covered in Fintechbits’ report on the Great Hill deal.
Fintechbits analysis
Why a Five-Year Term at a Regulated Carrier Matters
The term is the story in the Columbia Insurance Kalepa contract. Carriers rarely sign five years for software they have not tested, and Columbia describes a process long enough to move past the checklist stage. Kalepa gains a reference customer described as a small-business carrier rather than an E&S business, which helps when it pitches standard-market insurers. Rivals selling underwriting workbenches have to match a vendor that has a named carrier on a long contract.
The risk is on Columbia’s side. A multi-year commitment fixes its operating model before it can prove the premium and loss ratio gains it set as the target. Columbia has given no figure to hold it to. If outcomes lag, switching costs will be high.
For customers of Columbia, the practical change is likely to be speed and consistency on small-business submissions rather than different coverage. For regulators, it adds another carrier whose underwriting decisions will need an audit trail that explains where AI acted.
Frequently asked questions
What is the Columbia Insurance Kalepa agreement?
The Columbia Insurance Kalepa agreement is a five-year partnership under which Columbia uses Kalepa’s AI platform for underwriting and back-office work. It was announced on 17 November 2026 without financial terms.
What will Kalepa’s platform do under the Columbia Insurance Kalepa agreement?
Kalepa’s platform automates submission intake and clearance and supports triage, risk selection, rating and portfolio management. Columbia’s underwriters are meant to spend more time on complex risks and agent relationships.
How does Columbia plan to measure results?
Columbia’s chief underwriting officer framed the test as gross written premium growth and an improved loss ratio. The Columbia Insurance Kalepa announcement gives no target figures or review dates, so outside observers have nothing to check the outcome against.
What to Watch Next
The first marker for the Columbia Insurance Kalepa deal is any disclosure of scale. Columbia has not said how much premium will flow through the platform or when underwriters begin using it, and annual statutory filings would show whether growth follows.
The second is whether Kalepa’s newsroom adds a fourth carrier in 2026, and whether any other announcement states a term. The third is the governance detail: Columbia refers to consulting and systems integration partners without naming them, and an independent description of how AI decisions are logged would answer the traceability question Wharton raised.



