INSURETHEON-CHAINAGENTS
Insurance for the AI agents you trust with money. If an agent is tricked into breaking its rules, you're paid back on-chain the same day, with no adjuster and no lawsuit.
ENSv2World ID
Uniswap v4Paid by evidence, never by argument.
Rules the agent can't argue with.
The agent's ENS name publishes its rules: a per-transaction cap and a counterparty allowlist. A Uniswap v4 hook reverts any swap that breaks them. The Grok-style attack never executes.
Violations anyone can recompute.
A transfer that slips through is recorded on-chain. Over the cap or off the allowlist is arithmetic on public data. No oracle, no adjuster, re-run the check yourself.
A fresh human, then the money.
The policyholder re-verifies with World ID: the same human who bought it, right now. The shared reserve pays in minutes. Cancel, and the claim is held, never paid.
The losses standard policies stopped covering.
$150K gone to one message.
An attacker gifted Grok's Bankr wallet an NFT that unlocked elevated permissions, then a Morse-coded message talked the agent into sending roughly $150–200K. A safety block had stopped the same trick before. It didn't survive a code rewrite.
A premium you can read.
A credibility-weighted claim frequency, a tier from how tight the rules are, a discount for a clean streak. Every term computed on screen, nothing guessed.
ENSv2No Sybil farming the pool.
Before buying, the policyholder proves they're a unique human. One person can't spin up a hundred wallets and a hundred claims against the shared reserve.
World IDEnforce where you can. Insure what gets through.
Swaps through the pool are blocked before they execute. Transfers that break the rules are recorded, and a verified claim is paid from the reserve in the same afternoon.
Uniswap v4