Reinventing insurance
from the claims up

Insurance is a pool. A group of people who face the same kind of risk each pay a small amount into that pool. Most of them will never need to draw from it. The few who do get paid out of the premiums the majority contributed. The math is simple: spread the cost of an unlikely event across many people, so that no single person bears the full impact.

The insurer is the risk taker. They collect the premiums, they back the pool, and they pay the claims. In most periods, the premiums collected exceed the claims paid, and the spread is their profit. In a bad period, like a hurricane or a surge of claims, claims exceed premiums and they eat the loss. That spread between premiums and claims is the yield, and the risk of a bad period is why the yield exists.

This is not complicated, and it is not new. It is how Lloyd's of London has worked since the 1600s, how mutual insurance has worked for longer, and how every insurer works today. The structure is sound. What is broken is everything around it.

When you buy insurance today, you are buying a promise written in legal language, priced by an actuary you will never meet, and honored, or not, by a claims officer who answers to a balance sheet rather than to you. Some facts are binary and verifiable: a flight was delayed, a crop failed, a shipment arrived late. Whether you get paid for any of them is still a negotiation. The industry has spent a century optimizing the negotiation and almost no time questioning why there is one at all.

The opacity is not incidental; it is the business model. You cannot see the loss ratio on the policy you just bought. You cannot tell whether the premium reflects your actual risk or the insurer's quarterly target. A denial arrives as a form letter citing a clause you cannot independently evaluate. Weeks pass. Paperwork multiplies. Most claimants give up, which is, from the insurer's perspective, the system working as designed.

Here is the structural problem: the insurer is both the risk taker and the claims decider. They collect your premium, they hold the money, and they decide whether to pay your claim. Every denied claim improves their bottom line. That is not a conspiracy. It is the incentive structure. The policyholder and the insurer are on opposite sides of a transaction where only one side gets to judge whether the transaction completed.

This is the trust problem. Not that insurers are dishonest, but that the system gives you no way to verify anything. You trust the price is fair. You trust the reserves are adequate. You trust the denial was honest, even though a denial saves the denier money. You have no alternative, because the books are closed and the judge is also the defendant.

The conflict of interest, where the insurer decides whether to pay the claim while keeping the money if they don't, has a fix. Separate the risk taker from the claims decider. Let the insurer be the pool, but let the claim be settled by code that neither party controls.

That code is the oracle, a piece of the Insur technology stack. The oracle reads the truth and triggers the payout directly. For parametric coverage, where the trigger is an objective fact, it reads external data feeds: the flight landed 73 minutes late, the policy triggers at 60, the payout fires. For indemnity coverage, where the claim requires evidence and judgment, it reads encrypted evidence and a bonded committee's verdict. There is no claims officer in the loop, no discretion to exercise, no balance sheet to protect. The oracle is the decider, and it has no money to save by saying no.

This is what makes insurance trustless. The blockchain, the token, and the pool are the rails the oracle runs on. The oracle is the part that changes who you are trusting. Instead of a company that profits from your denial, you are trusting code that runs the same way for every claim and every policyholder, with every decision recorded on-chain and auditable by anyone.

The hardest part of insurance for the customer is not the price or the coverage. It is the claims process, the part policyholders hate, and it exists because a human has to look at each claim and decide whether to pay it.

Parametric coverage is the simplest case. The trigger is an objective fact that multiple sources agree on within minutes: a flight landed late, a temperature crossed a threshold, a sensor reported a flood level. The oracle reads the data feed and settles the moment the trigger confirms. There is nothing to review because there is nothing subjective: no form to file, no adjuster to convince, no waiting period. The only reason these claims still pass through a human today is that the rails were built in 1970 and never replaced.

Indemnity coverage is harder. The claim depends on evidence: a receipt, a damage report, a lost-bag filing. That evidence is digital and timestamped, but it requires judgment to evaluate. The oracle routes it to a bonded committee that decrypts the evidence, votes, and settles in hours rather than weeks. Bad-faith assertions get slashed. Every decision is a public record with a reason attached, and the policyholder can watch the whole process unfold on-chain.

For parametric coverage, the claims process disappears entirely. For everything else, it becomes a transparent, auditable flow instead of a black box. The friction was never necessary. It was a byproduct of the rails.

In traditional insurance, the premiums you pay disappear into a corporate balance sheet. When the year goes well and there are no hurricanes or surges, the profits flow to shareholders. The people who bought the coverage see nothing. The insurer keeps the spread, and the math behind it stays closed.

Insur opens capital participation in the vault. Anyone can supply liquidity to the vault's LP pool and become an LP. When a policyholder pays a premium, 97.5% goes into the vault's premium pool that backs active policies, and 2.5% goes to the revenue pool for INSUR stakers (protocol fee). Both the premium pool and the LP liquidity behind it are deployed into a lending protocol, where the capital earns yield while it waits. That lending yield, compounded back into the vault, is what LPs earn, on the premium pool and on their own deposit alike. The premium pool itself is the first funds drained when a claim hits; LP principal is only touched when the premium pool runs out.

INSUR token holders govern the protocol and earn a share of its revenue, a fixed 2.5% protocol fee on every premium, distributed on-chain in stablecoins. They do not take insurance risk; they own the protocol that routes the capital, and their return comes from that ownership. LPs bear the underwriting risk and earn the lending yield on deployed capital. The flows are auditable, and neither layer is a corporate parent skimming the float.

This is what Aave did to savings accounts. It did not win by being safer than a bank; the FDIC exists. It won by making every deposit, every borrow, and every interest payment a public transaction, and by passing the revenue to the depositors instead of the shareholders. The moat was not the yield rate but the accountability. Insurance is the same opportunity at a larger scale, with a deeper trust deficit.

A trustless oracle and an open capital pool are not a product. They are the rails every verifiable coverage will run on. The reference vault starts on travel parametric, where the trigger is objective and the claims process is most broken. The solvency check, the bind, the oracle settlement, the payout order, the fee split: every category after that reuses them. A new product is a new data source and trigger condition, not a new protocol.

This is why the protocol is vaults, not one pool. A vault is a scoped risk surface: one category, one curator, one claims mode, its own LP capital. The reference vault is the calibration anchor; a governance-listed curator runs another vault on the same rails. The underwriting and capital backend is reusable, which is how one travel-parametric vault becomes the infrastructure for the category.

The openness is the lever, and it runs one way. Today an insurer can keep its loss ratio, its denial rate, and its pricing logic in a black box, because no one else is publishing theirs. The moment a verifiable alternative exists, the box is a liability. A policyholder who can compare a public payout record against a form letter stops accepting the form letter. An LP who can read real loss ratios stops taking an insurer's word for them. Insur does not need to replace every carrier to change how every carrier behaves. It has to exist, publicly, as the baseline they are measured against.

That is the opportunity. Verifiable claims, open capital, and an oracle that cannot be lobbied are a new model for insurance and a new floor for everyone else. The incumbents get more accountable not because they choose to, but because the rails make the old opacity impossible to defend.