The User Experience Is the Tip of the Iceberg. The Product Is the Claim.

Dan Bratshpis co-founded INSHUR around 2017 in New York, starting with professional black car drivers — a deliberately narrow beachhead that turned out to teach the company everything. When I first met his co-founder David Daiches back in 2017 or 2018, I was still at Farmers, and what struck me was that INSHUR had actually understood its customer. If you insure black car drivers, your customer spends their working life inside a vehicle. Every interaction has to run through a phone. INSHUR built the entire onboarding that way, and the insurance product followed from the behavior rather than the other way around.

Eight years on, the company is roughly 250 people across New York, the UK, Westlake Village in California, and a claims operation in Dallas, operating in the US, the UK, and mainland Europe. Dan’s co-founder David is based in Brighton, south of London.

The thesis has stayed constant: the world is moving on demand. Food, rides, packages, and soon vehicles themselves. What most people miss is how much the on-demand economy depends on insurance — these platforms spend enormous energy and capital making sure risk is handled. INSHUR sells B2B2C, embedded into partners including Uber and Amazon.

In Episode 138 of InsurTechTalk, Dan and I covered why INSHUR built its own claims operation, how COVID nearly killed the company and accidentally saved it, and why the coming fight over autonomous vehicle liability is a rerun of a movie he has already watched once.

About Dan Bratshpis

Dan Bratshpis is the CEO and co-founder of INSHUR, an embedded insurance company built for the on-demand economy, covering ride-share, last-mile delivery, and car-share risks across the US, UK, and mainland Europe. INSHUR operates as an MGA and has built its own third-party claims administration on both sides of the Atlantic, since spun out under its own brand, Komodo Claims. Dan is based in New York; his co-founder David Daiches is in the UK.

Claims Is the Product

The sharpest reframe in the conversation was about where INSHUR’s attention went after the early years.

• Early InsurTech, by Dan’s own account, focused heavily on user experience — and INSHUR started there too
• But what you are ultimately selling is a piece of paper promising to pay a claim; protection when something goes wrong
• Around 60% of every premium dollar is spent on claims, making it by far the biggest lever available
• INSHUR could not find a claims solution that fit, so it built a bottoms-up TPA in both the US and Europe
• Dan’s framing: user experience is the tip of the iceberg — the substance sits below it in how the policy is priced, which rating variables are used, and how claims actually get handled

What Was Broken in the TPA Model

Dan was specific about why the existing third-party administrator market did not work for a business like his.

• Incentives cannot be aligned. Most TPAs are compensated by the number of checks written, so the more payments that go out the door, the more fees they generate
• You cannot A/B test. TPA selection is usually made alongside capacity partners, often on strong recommendation, and everyone promises great service and good pricing — with no practical way to compare one against another
• Pricing design is a menu. A bodily injury claim costs X, a property damage claim costs Y, with little innovation beyond that. INSHUR also experimented with percentage-of-premium arrangements
• Scale is a genuine constraint. INSHUR sits in a high-frequency, low-severity line — its customers are on the road roughly ten times more than an average consumer, producing claim volumes most TPAs are not equipped for
• Dan’s alternative: reward an adjuster for handling a claim well and saving money, not for the volume of payments processed

That operation has now been rebranded as its own company, Komodo Claims — a play on commercial auto — and is being offered to outside parties.

COVID: The Pivot That Became a Blessing in Disguise

INSHUR’s business essentially ground to a halt in 2020. Nobody was taking rides.

• Existing customers began doing delivery work instead — Uber Eats, Deliveroo, DoorDash, Amazon Flex, Wolt — and came asking whether INSHUR had a product for them. The answer at the time was no
• The company pivoted into last-mile delivery, both food and parcel, which is now a major part of the business
• That pivot produced the usage-based wallet, developed in the UK: rather than buying an annual policy costing thousands, a courier deposits a small amount — around £25 — into a wallet. INSHUR plugs into the delivery platform, draws a predetermined premium while the courier is working, and switches off when they stop, covering only commercial use
• A third pillar followed in car share: hundreds of thousands of US vehicle owners rent out cars on platforms like Turo. INSHUR’s Period X product covers the host while the vehicle is in their care, custody, and control, with the platform’s policy taking over during the rental

INSHUR does not currently insure scooters, though it covers a significant volume of two-wheel delivery couriers.

Autonomous Vehicles: Watching the Same Movie Twice

Dan’s déjà vu is specific and worth taking seriously, because he lived through the first iteration.

What happened last time: around 2010–2012, as Uber, Lyft, and Sidecar launched, there was no framework for how insurance would work. The platforms were nearly shut down — Lyft and Uber received cease-and-desist letters from Los Angeles. It took roughly a decade for a patchwork of state-by-state laws and regulations to settle.

What is happening now: California, Arizona, and Texas are each taking different approaches to autonomous vehicles. The unresolved questions are structural:

• Who carries the insurance — the OEM, the vehicle owner, the driver, or the company dispatching the vehicle?
• Which policy responds when something goes wrong? A conventional auto policy assumes one driver, one vehicle, one policy. An autonomous manufacturer carries auto liability, product liability, and potentially cyber. When an autonomous vehicle hits a pedestrian, or another autonomous vehicle, which of those responds?
• What happens in a handover scenario, where a human can override the autonomous system?

The genuinely good news, and Dan led with it: accident frequency should fall. Auto accidents are a leading cause of death, so fewer of them is a win for humanity regardless of how the insurance shakes out. The complexity sits in the middle period, where humans and autonomous vehicles share the road.

He also flagged a subtler shift in claims economics: when the vehicle involved carries the logo of a trillion-dollar company, the claims handling expertise required changes entirely. Deep pockets attract different lawyers and different expectations about what a settlement should look like — which cuts against insurance’s core purpose of restoring someone to where they were rather than delivering a windfall.

The Reinsurance Problem With Genuinely New Risk

Dan’s read on capacity was candid about how the market has changed since INSHUR started.

• Between roughly 2016 and 2018, reinsurance was abundant and, in his words, a lot of InsurTechs got very lucky — reinsurers were actively seeking new risk and several ran corporate venture arms funding the sector
• The bar today is materially higher: capacity partners expect real underwriting returns from whoever handles pricing, underwriting, and claims. Dan considers this a healthy correction
• His summary of the old mindset: how much direct written premium can I write, and can I count it as ARR
• The structural challenge for autonomous risk is that reinsurers are risk-averse by design and need credible data to part with balance sheet — and for a genuinely new risk, that data barely exists
• INSHUR manages this by starting new programs slow and steady until performance data exists, taking some risk itself early on, and cross-subsidizing emerging programs with established ones

INSHUR operates as an MGA with sliding-scale arrangements with reinsurance partners and a small captive, but for the most part does not carry the risk itself.

The Closing Lesson: Unit Economics Before Growth

Asked for advice for emerging MGAs, Dan went straight at the numbers.

• There are a hundred pennies in a premium dollar, and most of them should go to claims payments — typically 40 to 70 cents
• When he sees a program running a 10% loss ratio, alarm bells go off, because something is structurally wrong
• The rest breaks down across claims administration, underwriting profit for the capacity partner carrying the risk, and the distributor who found and sold the deal
• Both the expense ratio and the loss ratio matter — and his blunt version: anybody can sell a dollar bill for 50 cents, the question is how you sell it for more than a dollar
• Get the unit economics right before growing quickly

Key Takeaways

• Claims consume roughly 60% of every premium dollar, which makes claims operations — not the app — the highest-leverage part of an insurance business
• The standard TPA model pays for check volume rather than outcomes, an incentive misalignment that is very hard to fix from the outside
• High-frequency, low-severity lines break most TPAs on scale alone, independent of incentives
• COVID destroyed INSHUR’s original market and forced the delivery pivot and usage-based wallet that now anchor the business
• Autonomous vehicle liability is replaying the 2010–2012 ride-share regulatory scramble, state by state, with the added complication of overlapping auto, product, and cyber policies
• Reinsurers need credible data to back new risk, which is precisely what genuinely new risk cannot provide — the central chicken-and-egg problem for insuring autonomy
• For emerging MGAs: fix unit economics before adding growth, and treat an implausibly low loss ratio as a warning rather than a win