Disability insurance claims have a reputation in the industry: tedious, document-heavy, resistant to shortcuts. An examiner might spend days reconciling fragmented medical records with employment histories and policy fine print. It's the kind of workflow that has historically made underwriters wince and technologists shrug.
John Haller believes he's found a way in.
His St. George-based startup, Curant.ai, announced in late July that it raised $3.1 million in seed funding to build AI infrastructure purpose-built for these arcane processes. Diagram led the round, with participation from Humania Insurance, Element Ventures, Blue Plains Capital, and a handful of angel investors whose names weren't disclosed.
The deal is notable less for its size than for its structure. Humania Insurance isn't just writing a check—it's actively using Curant's platform in production, a dual role that serves as both commercial validation and competitive moat. Strategic investors de-risking their bets by testing technology before funding it: it's a pattern emerging across enterprise AI, though not always a guarantee of durability.
When Your Investor Is Also Your Beta Tester
Luc Thibault, a senior vice president at the Canadian life and health insurer, described Curant as providing "a safe, secure intelligence layer" atop Humania's existing system of record. It's the kind of language that suggests careful internal vetting, perhaps more than the founders initially expected.
Humania has been unusually aggressive in adopting AI tooling. Just two months earlier, in May, the company announced a separate partnership with Koïos Intelligence focused on quality assurance and compliance automation. The implication? Humania isn't betting on a single vendor to solve its problems. It's assembling a portfolio of AI tools, each carving out a specific niche in its operations.
For Curant, that means proving its value in a competitive internal landscape—not just against manual processes, but against other automation efforts vying for the same budget and attention.
What the Platform Actually Does
Curant's system ingests complex claim files and queues, then processes them into what it calls "review-ready insights." Think: data extraction, fraud detection, timeline reconstruction, case recommendations. The human examiner remains in the loop, but ideally spends less time hunting for information and more time making decisions.
The company offers some internal pilot metrics: four hours saved per claim per associate, a 35 to 50 percent reduction in manual analysis work, and twice as many data fields captured compared to manual review. Those numbers haven't been independently verified—Curant acknowledges a formal partner case study is still being developed. Take them as directional rather than definitive.
Haller brings a somewhat unusual résumé to the problem. According to the company, he's spent over 15 years in insurance technology and holds credits as inventor or co-inventor on 17 patents. But before founding Curant in 2026, he co-founded beatBread, a music fintech venture that provided financing to artists—a detour that suggests comfort operating outside traditional industry boundaries. Now he's back in insurance, tackling one of its thorniest operational challenges.
A Crowded Moment for Insurance AI
The funding arrives amid a flurry of activity in insurance infrastructure. In recent months, Outmarket AI raised $17 million, Cara closed $8 million, and Corgi pulled in a $160 million Series B at a $1.3 billion valuation. Diagram, the lead investor in Curant's round, operates as both a venture builder and early-stage fund, backed by Sagard and focused primarily on fintech and climate tech.
Curant says the new capital will go toward expanding its AI platform, deepening carrier partnerships, and helping insurers shift from isolated automation pilots to what it characterizes as enterprise-wide transformation. That's a common pitch in the category, though the execution varies wildly depending on legacy system complexity and organizational readiness.
For now, Curant is targeting disability claims specifically. The company positions its platform as extensible to other complex insurance workflows down the line—workers' comp, long-term care, perhaps even underwriting. Whether that modularity holds up in practice, or whether each insurance line demands its own bespoke solution, remains unclear.
The Augmentation vs. Replacement Question
The broader question hanging over deals like this: Are we watching a genuine leap in claims operations, or just another layer of software destined to complicate rather than simplify?
Curant frames its offering as augmentation, not replacement—turning unstructured case files into structured decision support for human examiners. It's a safer narrative than promising full automation, and probably a more realistic one given regulatory scrutiny and the high stakes of disability determinations.
But augmentation can be a transitional phase. Today's decision support tool can become tomorrow's autonomous agent, especially as models improve and insurers grow more comfortable delegating judgment calls to algorithms. Haller and his team are building for the current reality while betting that reality will shift in their direction.
Whether disability claims prove to be the right wedge remains an open question. But for Humania, at least, the experiment is already underway.
