Remember when insurtech was going to tear up the insurance playbook? Around 2021, that was more or less the pitch, when insurance was slow, buried in paperwork and about as beloved as a trip to the dentist, so a wave of startups set out to rebuild it from scratch.
Investors loved the story. Global insurtech funding topped $15 billion USD that year, per CB Insights, but it didn’t quite go to plan. An industry built on capital reserves and state regulators, it turns out, doesn’t fold because someone shipped a nicer app.
Plenty of those startups are gone now, with the ones still standing mostly having learned something that sounds obvious in hindsight: you don’t have to drag customers to your product if you can put the product where they already are.
Now, money is flowing again. Gallagher Re’s latest Global InsurTech Report puts second-quarter funding at $2.44 billion USD, the most since 2022. Almost all of it, 99.1%, went to AI-focused companies, while early-stage funding dropped 51.8% from the quarter before.
“Capital availability is clearly not a problem,” Andrew Johnston, Gallagher Re’s global head of insurtech, said about the numbers. He called what he’s seeing a paradox; AI is supposed to be making everything cheaper, yet individual insurtechs are raising and then burning through more cash than ever.
So getting a meeting with an investor is easier than it’s been in years. Building something that lasts is another matter entirely.
Nobody goes shopping for insurance
Nobody spends a Saturday browsing renters insurance for fun. You buy it because a landlord or a lender tells you to, usually with a deadline attached.
Embedded insurance, meaning coverage offered inside some other purchase, works with that habit instead of fighting it. And it also sells better. BCG found that traditional insurers going this route are already seeing higher conversion rates than when they sell standalone coverage for the same products.
Cover Genius is probably the clearest proof that the model scales. Its platform connects more than 200 partners with over 50 insurance carriers and protects upwards of 70 million customers at the point of sale, according to FinTech Global.
Most of those customers never went looking for a policy. They bought a flight or booked a ride, with partners like Booking.com and Uber, and the coverage came with it. In July, the company raised $100 million at a $1.9 billion valuation. Notably, the money came from Vista Equity Partners’ credit arm, not from another venture round.
Regulation pushes in the same direction. Anyone who sells, solicits or negotiates insurance in the U.S. has to be licensed as a producer under state law, according to the National Association of Insurance Commissioners.
But most software platforms want nothing to do with that headache; insurers hold the licenses but rarely own the customer relationship and a startup that stitches the two together gets paid for solving both problems.
Rental housing is about as clean an example as you’ll find. If you’ve signed a lease recently, you know the drill: sign here, then show proof of renters insurance before you get your keys. Get Covered, a New York-based insurtech, built its business around that exact moment by plugging into the property management systems landlords already use, like Yardi and Entrata.
The company says it now powers insurance compliance for more than 3 million rental units.
CEO Brandon Tobman walked through how that works. The tenant signs in the property manager’s portal and lands straight in a flow to buy coverage or upload a policy they already have, no second website, hunting around for a login.
The embedded products that work best, he argued, feel like part of the service rather than an upsell.
If you’re building in a completely different space, steal the question anyway. Whose workflow are you living in? If the only honest answer is your own app, acquisition costs will quietly eat the business model, regardless of how good the product is.
There’s a less obvious point in Tobman’s argument that deserves more attention. While most people focus on the offer itself, most of the hard work happens before any customer sees one, when both sides sit down and match the data a platform already holds, like lease details, against the bare minimum an insurer needs for a quote. Get that right and the form shrinks. Fewer people bail halfway through.
Sure, it’s unglamorous work. But it’s also the moat. Anyone can copy an interface in a couple of sprints, but a dozen enterprise integrations that clients actually trust take years to rebuild.
Buying the workflow, not building it
Deal activity tells the same story. Real estate tech startups have raised about $8.7 billion so far in 2026, Crunchbase reports, a far cry from the $24 billion USD raised in 2019.
M&A has been busy, though, with incumbents buying data, workflow ownership and distribution so they can build credible AI products faster. Autodesk paid $3.6 billion USD for MaintainX, for example. CoStar spent $800 million USD on Zonda.
Insurance is seeing the same thing. In August, Munich Re agreed to acquire cyber insurtech At-Bay at an enterprise value of $575 million, with Munich Re’s HSB unit having been At-Bay’s strategic partner since the startup’s founding in 2017.
Scale-ups are buying their way into new channels, too. Cover Genius acquired Friendsurance in July, picking up a Berlin-based platform that already had the bank integrations and regional regulatory know-how to reach customers in Germany, Austria and Switzerland.
And in June, Get Covered acquired Revyse, a vendor compliance and contract management platform, and described the deal as the first step in an M&A strategy meant to speed up product development. Instead of spending a couple of years building vendor tools, it bought a team that had already done that work and plugged it into distribution it had spent years putting in place.
In both cases, the buyer skipped years of building and paid for a workflow someone else had already earned.
For founders, the read here is a little uncomfortable: own a workflow people rely on every day and you end up as either the buyer or the target, and both are decent outcomes; be a feature sitting on top of someone else’s system, and you’re the easiest thing in the stack to rip out.
AI still matters, of course. Investors are putting nearly every insurtech dollar there. But a model with no route to customers is just a demo with a burn rate.
The founders who get through this cycle in good shape will be the ones who worked out where their customers already spend their time, and built the pipes to reach them there.
Featured image: Mehdi Mirzaie via Unsplash+
