Insurance AI's test moves from the call to the claim

The raise

Kalepa, which builds AI-driven insurance workflows, raised $75 million in a Series B round led by Updata Partners, bringing its valuation to $470 million and its total funding to date to $91 million. The company says it plans to use the capital to expand its product suite, enter new markets and double its headcount in 2027. It also says it has tripled its revenue and customer count over the past year, that the volume of work running through its platform grew fourfold in the past year, and that it has been cash flow positive for more than two years. m:…6853

The numbers Kalepa's customers report are the ones a buyer would actually check. Insurers using Kalepa report a 58% reduction in quote time and a 960-basis-point improvement in combined ratio, according to the company. Those are underwriting economics, not demo metrics: quote time is a sales cost, and combined ratio is the difference between profit and loss in insurance. A $470 million valuation for a company claiming both growth and cash flow positivity is a bet that those two numbers travel together into new markets. m:…6853

From capability to execution

Liberate's argument this week is that the industry's question about AI has changed. The primary question has shifted from capability to execution, the company says, and success in 2027 will be defined by how well carriers build AI into the actual work of claims, sales and service. The specifics it points to are unglamorous: first-notice-of-loss actionability, intelligent handoffs and auditability. Carriers will need to demonstrate not just that AI completed a task, but the specific process and logic used. m:…8159

The regulatory backdrop is moving the same way. Liberate notes that existing state insurance laws apply to AI-driven decisions just as they do to human decisions, and that the NAIC began piloting an AI Systems Evaluation Tool with 12 states in 2026. The customer side is already digital: 38% of homeowners customers reported filing a claim digitally in 2026, while 21% of auto insurance customers were forced to switch channels to resolve a single inquiry. The demand is there; the handoff between channels is where it breaks. m:…8159

The claims floor

Five Sigma's post is a practical version of the same argument: automate the routine, high-frequency claims workflows first, and extend autonomy one workflow at a time based on evidence of accuracy. Its First approach has three stages — Assist, Collaborate and Lead — and adjusters retain control over coverage determinations and denials. The customer results it cites are concrete: INSHUR UK reduced claim email handling time by 60% and achieved claim-matching precision of about 97%, while Veygo by Admiral saw a 60% reduction in first-notice-of-loss set-up time. Resorts World Las Vegas reviews documents with hundreds of pages in 37 seconds. m:…4230

The logic is that routine tasks account for most of the idle periods in a claim's lifecycle, so automating them shortens the claim even before any single task gets faster. Agentic AI, in Five Sigma's framing, interprets the claim as it arrives and determines which task applies, rather than executing a predefined instruction when a structured field meets a condition. Escalation criteria are defined in advance from the insurer's standard operating procedures, which is the auditability Liberate is asking for. m:…4230

The harder market next door

Mortgage AI shows the other side of the same coin: execution matters, but the market sets the ceiling. 21X, which builds AI agents for lenders, notes the cost to produce a single mortgage loan is approximately $10,900, and that 30-year mortgage rates are holding above 7 percent while the Mortgage Bankers Association expects origination volume to continue falling into 2027. Lenders are increasingly focusing on database reactivation of past clients and aged leads because refinance incentives are low. m:…7858

The pitch is that AI agents differ from chatbots by taking active steps on a loan file rather than just answering questions: they can call and text leads immediately upon arrival in the CRM, and 21X performs live transfers of qualified borrowers with full conversation context attached. The first lender to contact a rate-shopping borrower is usually the one that secures the application, according to the company. In a shrinking market, speed of contact is the margin. In insurance, the margin is the claim. Both point the same way: AI that does the work, not AI that answers the phone. m:…7858