About
Everything you'd
ever pay for.
For a sip.
A sip is a hundredth of a cent - the smallest useful unit of payment, finally cleared by stablecoins after thirty years of being blocked by credit-card fees. The Sippar story is what becomes possible once that unit exists: the subscription stack collapses, services price themselves per use, and an agent pays on your behalf with a key no operator can sign for, including us. We are the identity layer for the part of the economy that runs without a human in the loop, so that the part that does run with one stops being billed monthly for things it doesn't use.
The one residue: until the signer canister's controller closes, Sippar could still redeploy that code. That's why the honest label today is "Sippar-managed, on a scoped path to non-custody" - not the finished state, just the one we're building toward.
Why this name
Sippar was a real city.
Sippar was continuously occupied on the bend of the Euphrates for roughly 2,400 years across seven empires. Its temple housed the cult of Shamash - the Mesopotamian sun god of justice, whose defining property was that he saw everything that happened in the world every day. The Code of Hammurabi was likely erected there. The oldest known map of the world was made there. The temple library, excavated intact in 1986, sat physically between the courtyards of justice and witness.
We named this project after that city because the work it did - verification between observers, connective infrastructure across rising and falling polities, a coordination layer at the center of a settlement topology - is the shape of what we are building.
Who
Built by one founder, on purpose.
Sippar is built by Elad Mintzer, a solo founder in Berlin, writing code and shipping daily.
The team is small on purpose; the architecture has to be defensible without a 30-person ops org standing behind it.
What
The relay, not a bridge.
Your agent holds funds on one chain; the service it needs lives on another. Sippar closes the gap - the agent pays from its own wallet on its home chain, Sippar pays the service on whichever chain and rail the merchant accepts, and each leg settles on its own chain. No bridge, no wrapped tokens.
The key is split across 34 nodes on ICP's fiduciary subnet; no single party can sign with it, no single party can be coerced into signing with it, and the private key, in the conventional sense, never gets assembled anywhere. Including by us.
The same identity is designed to reach card-network agent-payment protocols as they mature - Visa's Trusted Agent Protocol, Mastercard's agent toolkit, Google AP2 - alongside Tempo MPP today.
The model
The Wise of agent payments.
Wise rebuilt international money transfer without making anyone learn how wire transfers work. Wiz made cloud security usable by people who never wanted to become security engineers. The pattern is always the same: take an infrastructure layer that used to demand expertise, hide the expertise, expose a clean interface.
Sippar does this for agent payments. Underneath, your agent's funds move across every chain we support, signed by keys split across 34 independent operators and never assembled in one place. On top, your agent just pays for what it needs and you read one receipt. The chains, the signatures, the settlement: ours to handle, not yours to learn.
Why
Stripe's floor is the whole problem.
The credit-card system has a 30¢-plus-2.9% per-transaction floor that hasn't moved in fifteen years. That floor is the reason the web has subscriptions instead of price tags - anything under roughly $1 is economically broken on a card.
Stablecoins finally cleared the floor: USDC on Solana settles for about $0.0007. Sub-cent payments are not an upgrade to a card; they are a different category of payment, and they need a different kind of identity to authorize them.
The sip is the unit that makes "everything you'd ever pay for" a price tag rather than a marketing claim.
Built on
Production-proven primitives.
ICP Chain Fusion
Threshold signatures (t-ECDSA + t-Schnorr Ed25519)
Algorand
Instant BFT finality, ASIF agentic-commerce stack
Solana
pay.sh / Machine Payments Protocol
EVM
Base, Arbitrum, Optimism, Polygon, BNB, Ethereum
Stellar
~5s finality, lowest fees
TON
V5R1 wallet, STON.fi swap
What we won't trade
Five rules.
In the order a decision usually hits them.
Humans in charge.
The agent is the human's, never the platform's. Budgets are enforced by the canister, not by us. Identity is portable across rails and merchants. If we vanish tomorrow, your agent's address survives.
Pay for what you use.
Sub-cent settlement means a three-sip query costs three sips. No subscriptions. No bundles disguised as access. The receipt shows every charge with the USD breakdown, not a monthly aggregate.
Mathematical trust.
Security comes from cryptography, not from trusting an operator - including us. Threshold signatures instead of custodial keys. DePIN substrate instead of cloud-provider trust. On-chain audit trails instead of off-chain databases.
Honest architecture.
What we ship is what we say. The gaps are named publicly. The trade-offs are not buried in footnotes. The fact-check checklist is a feature, not a liability.
Universal access.
Expert AI at the price of a single useful query, not at the price of a subscription nobody fully consumes. The cheapest user is the design target, not the highest-revenue one.
Honest scale
Mainnet only. Real transactions. Real users. Real value.
What's still small
Three gaps, named up front.
There are gaps. Three worth naming up front.
The market is small.
Daily transaction volume across the protocol today is in the tens of thousands of dollars, growing from a smaller base. That's normal for a new payment rail in its first year. Every rail launches small. The thesis bets on the trajectory, not the current state.
The law hasn't caught up.
No major jurisdiction has settled how money-transmission rules apply to keys held by signing schemes rather than companies. The cryptographic case is defensible. The legal case is in motion. Regulatory clarity will take time.
The infrastructure isn't done.
The signing layer underneath agent payments is in production for one major protocol family (EVM and Solana-style chains) and Tempo MPP. Card-network protocols - Visa's Trusted Agent Protocol, Mastercard's agent toolkit - are real external standards it's designed to reach as they mature, not live yet. Google's AP2 uses an elliptic curve the current signer doesn't yet expose. The dependency on the operator running it today is real, and removing it is on the roadmap rather than shipped.
Contact
Tell us what you're building.
Partnerships, integrations, grants, or just curiosity - we read every message and reply within a day. What you send lands in the founder's inbox, not a CRM workflow.
Or reach us directly: