How RiskCube Built AI to Fix American Insurance

Startup founders need insurance to protect against lawsuits, data breaches, and executive liability claims, but the process of finding the right coverage remains frustratingly opaque. Each carrier requires its own application; the same company can receive quotes ranging from $6,000 to $50,000 for identical coverage with no explanation for the variance, and critical exclusions hide in contract language that would take a lawyer hours to parse.

Andrei Craciunescu founded RiskCube to address this problem by bringing enterprise-grade risk analysis to venture-backed companies. The brokerage platform uses AI to decode policy documents and compare quotes across the market, giving founders the transparency they need to make informed decisions.

A System Built to Confuse

Insurance protects companies from lawsuits, data breaches, and executive liability claims that could otherwise end a business overnight. A single cyber incident or a directors and officers (D&O) claim can cost millions, which makes adequate coverage a must-have for any company, especially those guarding private data or taking outside investment. Yet finding the right policy at the right price remains remarkably difficult.

The challenge stems partly from the contracts themselves. Insurance policies are typically dense, lengthy, and near-impenetrable legal documents, in which critical exclusions in coverage can be easily hidden in ways that many founders (especially new ones) lack the expertise to decode.

Startup founders looking for insurers also typically deal with incompatible systems, as each carrier has its own separate application form. The same risk can generate wildly different quotes ($6,000 from one carrier, $10,000 from another, $50,000 from a third) with no explanation for the variance, leaving startups to collect quotes, read policies, and conduct their own analysis in a process that can take hours. Existing insurtech platforms have optimized for speed, delivering fast quotes from a single carrier, but speed without comparison leaves founders unable to verify whether they are overpaying or underinsured.

The result is the equivalent of buying a car from the first dealership visited simply because the salesperson was quick.

Andrei Craciunescu, founder of AI-native brokerage RiskCube, believed the industry needed a fundamentally different approach. "All these carriers at the moment who offer startup insurance," he explains, "they give you one quote. But you don't know if you're overpaying or what is included or excluded."

RiskCube: Risk Quantification Meets the Brokerage Model

Craciunescu spent nearly a decade working across every segment of the insurance value chain: insurer, reinsurer, and broker. He began his career at a life insurance company before moving to Munich Re, one of the world's largest reinsurers. He later joined Willis Towers Watson, where he advised Fortune 500 CFOs on risk quantification while pursuing doctoral work at the Technical University of Munich's Department of Risk and Insurance.

That combination of practical knowledge and academic research was what Craciunescu leveraged to create RiskCube. The system uses natural language processing and LLMs to scan policy documents, flag exclusions and coverage gaps, and model scenarios in which claims would likely be denied.

For example, consider a seed-stage startup looking to buy four different insurance products: directors and officers liability, cyber coverage, tech errors and omissions, and general liability. In a typical scenario, this would mean four separate applications to multiple carriers, dozens of forms, and weeks of back-and-forth.

With RiskCube, the founder completes a single questionnaire that can be used to request quotes across all four insurance lines at once. That data is then distributed across the entire insurance market, and quotes return in a transparent comparison interface similar to Hotels.com or Amazon's product listings.

Once it's got all quotes lined up, the platform applies a total cost of risk calculation that takes into account the premium price but also any deductibles, coverage limits, exclusion clauses, and the possibility of claim denial to determine which policy delivers the best return on investment (ROI). Through this system, work that traditionally took brokers an average of ten hours to complete can now be performed in a matter of minutes and far more accurately

This approach also corrects a fundamental information imbalance. Carriers have far more pricing data than any individual startup, which means founders typically negotiate blind. RiskCube benchmarks pricing across the market, giving startups the leverage to negotiate and choose coverage with confidence rather than guesswork.

An Approach Validated By Its Early Traction

RiskCube's first transaction delivered proof of concept: a Y Combinator-backed startup with seed funding saved 70% on its premium simply through transparent comparison. The company has since sold directors and officers liability, cyber insurance, and tech errors and omissions coverage to YC-backed startups averaging $3 million in funding.

Building on this early traction, the company's team of engineers and designers is finalizing technology for a February launch on the Y Combinator platform. RiskCube also partners with specialty carriers that handle emerging industries, including AI, Robotics, Web3, space, and fintech, taking coverage that not all traditional carriers focus on and making it accessible to startups with restricted budgets. RiskCube already holds an insurance license in California and Delaware, with licensing expansion planned next for New York and Texas, two major hubs for startup activity.

RiskCube's approach has also drawn attention across the insurance ecosystem. Craciunescu was recently interviewed by Sabine VanderLinden, CEO of venture lab Alchemy Crew and a recognized expert in insurtech innovation, for the podcast Scouting for Growth at one of the largest U.S. insurance conferences, showing growing interest in alternative models for serving the next generation of businesses.

Whether the broader industry follows, Craciunescu notes, is still something to be determined, as the insurance commission structure creates systematic resistance to tools that lower premiums: brokers earn a percentage of what customers pay. When he attempted to sell risk analytics software to traditional brokers earlier in his career, he encountered consistent pushback, since adopting the tool would mean their revenue could be immediately reduced.

"The broker is not incentivized to use these tools because if they use them, they make less money because the premium might go lower," he points out. "So, why should they do this?"

RiskCube sidesteps this by owning the customer relationship directly, making sure company incentives are in tune with customer outcomes rather than premium maximization. This structure means the company grows by drawing in more customers rather than by steering existing ones toward higher-priced policies. As more startups look for transparency when choosing the best insurance for them, the traditional brokerage model may be looking at a new way forward.

An AI Fix for Broken Insurance

RiskCube represents Andrei Craciunescu's bet that startups deserve the same caliber of risk analysis that Fortune 500 companies receive. By embedding analytics directly into a brokerage without focusing on selling them to incumbents with misaligned incentives, the company seeks to offer new founders a way to compare policies transparently and understand what they are actually buying.