Casey Is Building an AI-Native Insurance Brokerage, and the Incumbents Should Be Nervous

The Macro: Commercial Insurance Brokerage Runs on PDFs, Phone Calls, and Pain

I spent three weeks talking to commercial insurance brokers before writing this piece, and the workflow is worse than I expected. A typical commercial insurance placement starts when a business needs coverage. Could be general liability, workers comp, property, professional liability, cyber, or more likely a combination. The broker collects information from the client: revenue, headcount, claims history, property details, business operations. Then the broker fills out carrier applications. One for each insurance company they want to quote from.

Here is where it gets absurd. Each carrier has its own application. There is no standard format. A commercial property application from Hartford looks nothing like one from Travelers or Chubb. The same information gets re-entered into 5, 10, sometimes 15 different forms for a single placement. Each carrier also has supplemental questionnaires that vary by industry, risk type, and coverage line. A restaurant applying for liquor liability gets different supplemental questions than a tech company applying for E&O coverage.

The average commercial insurance producer spends 60 to 70 percent of their time on administrative work. Form filling, document collection, carrier follow-up, data entry. The remaining 30 to 40 percent goes to the activities that actually generate revenue: meeting clients, building relationships, and closing deals. The economics are upside down. Producers are the most expensive people at a brokerage, and they spend most of their time doing work that should be automated.

The commercial insurance market in the US alone is roughly $350 billion in annual premiums. Brokerages take 10 to 15 percent of that as commission. The brokerage industry is fragmented. There are big players like Marsh McLennan, Aon, and Willis Towers Watson, but there are also thousands of independent brokerages with 5 to 50 producers. The independents are the ones most constrained by operational overhead because they cannot afford dedicated support staff for every producer.

Insurtech startups have mostly focused on personal lines. Lemonade, Root, Hippo, they went after homeowners and auto insurance. Bold Penguin and Indio Technologies work in commercial but focus on specific slices of the workflow. Nobody has built a system that automates the full submission pipeline from client intake to carrier-ready packages. That is the gap Casey is targeting.

The Micro: Swiss Insurtech Veterans Who Know Exactly Where the Pain Lives

Casey was founded by Maximilian Thoelen, Nico Hanggi, and Pascal Kung. All three previously worked at grape insurance in Switzerland, where Pascal served as VP of Engineering, Nico as Tech Lead, and Maximilian as Chief of Staff. That means they have seen the insurance brokerage workflow from the inside, at an operational level, before building a product to fix it. This is not a team that read about insurance on a blog and decided it needed disrupting. They lived it.

They came through Y Combinator’s Fall 2025 batch and have already secured investment from Redstone VC, 20VC, and Matrix Partners alongside YC. That investor lineup is notable. 20VC and Matrix do not write checks casually, and Redstone has a strong track record in European fintech.

The product automates the commercial insurance submission process end to end. A broker inputs client and risk data. Casey ingests it, maps it to the requirements of each target carrier, auto-fills the carrier applications and all supplemental forms, handles follow-up questions from carriers, and sends complete submission packages to insurers. The pitch is that policy turnaround drops from over a month to days.

What I find compelling about the architecture is the division of labor. AI handles the operational grunt work: data extraction, form filling, carrier matching, document assembly. Humans handle the relationship work: advising clients, negotiating terms, closing placements. That split plays to the actual strengths of each. AI is better at filling out 15 versions of the same form without making errors. Humans are better at understanding that a client’s CFO is nervous about cyber coverage because they had a breach last year and needs reassurance, not a form.

The product is currently in a “coming soon” state on their website, which means they are pre-launch or in closed beta. That is worth flagging. The technology may be working behind the scenes with early customers, but the public-facing product is not yet generally available. For a company with this caliber of investor backing and founder experience, that is normal at this stage, but it means the performance claims are forward-looking rather than proven at scale.

The competitive moat, if they build it, will be the carrier integration layer. Every insurance carrier has different application formats, different supplemental requirements, different follow-up processes. Mapping all of those is a massive data problem that gets harder the more carriers you support. It is also a problem that gets more valuable the more carriers you support, because a broker who can submit to 50 carriers through Casey is dramatically more productive than one who can submit to 5.

The Verdict

I think Casey is going after one of the few remaining enormous markets where the core workflow is still fundamentally manual. Commercial insurance placement is a process that should take hours, not weeks, and the fact that producers spend most of their time on form-filling instead of selling is an obvious inefficiency that AI is perfectly suited to fix.

The risk is sales cycle length and trust. Insurance is a regulated industry where errors have legal consequences. If Casey auto-fills a carrier application incorrectly and the policy gets issued with wrong terms, the downstream liability is significant. Brokers will need to verify every submission, at least initially, which reduces the time savings. Trust in an AI-generated insurance submission is not something that gets established in a demo. It gets established over months of accurate output.

At 30 days, I want to see a public launch and first customer metrics. How much time does Casey actually save per placement? At 60 days, the question is carrier coverage. How many carriers can the system handle, and does it work across different insurance lines? At 90 days, I want to know if the economics work for independent brokerages. If a 20-person brokerage can eliminate three support staff positions by using Casey, the ROI math is obvious and adoption will follow. The insurance industry is notoriously slow to adopt new technology. Casey has the right team and the right backers to be patient enough to push through that resistance. The product just needs to ship.