Confiance

XRPL native lending is the building block for on-chain credit. Single-asset vaults (XLS-65) let lenders pool funds, and the lending protocol (XLS-66) lets a loan broker issue fixed-term, uncollateralized loans from that pool, backed by first-loss cover the broker posts to protect lenders. Like real-world credit, this opens XRPL to borrowers without large crypto balances. But the protocol only moves funds; it does not decide who should get a loan, and every bad approval is a direct loss for lenders. A lender must never lend to a sanctioned wallet, needs a defensible reason for every yes or no, and must react quickly when a borrower falls behind. Today, a broker on XRPL Lending has to build all of this themselves.

Confiance (French for “trust”) is the decision layer on top of XRPL lending: an underwriting desk with three screens with their own roles. Lenders deposit into a private XRPL vault during its deposit window and earn the interest borrowers pay. A borrower picks a wallet, an amount and a number of payments, and authorizes a soft credit check. Seconds later they get a score, a rate, a level that sets how much and how long they can borrow, and a plain-language explanation. Once the loan exists, they get a score passport in their wallet and can repay in one click. The admin approves each recommendation, creating a real loan signed by both parties, and manages first-loss cover, tracks revenue, sees overdue alerts, impairs or defaults loans, pauses the agent, and sets the minimum score.

Behind each decision is an underwriting agent running a fixed pipeline: a sanctions gate against the official OFAC list, then a ban list of past defaulters, then consent. Only with consent does it pull the credit summary and read the wallet’s ledger history, combining them into the Ledger Credit Score (LCS): 60% credit file, 40% on-chain (account age, activity, trust lines, a KYC credential, and loans repaid with us). The score sets level A, B, or C, each with a maximum amount and term, and the rate falls as the score rises. Every offer is capped by what the vault and first-loss cover can support.

The rules decide, the AI explains, and a person approves. Every decision comes from plain, tested code; the explanation never changes it, and no loan is created without admin approval. Every decision is also provable: the full decision (score, reasons, rate, level) is hashed into the memo of the loan’s own transaction. Anyone can open the loan on the XRPL explorer and confirm it was made on exactly the approved decision; the borrower screen checks this automatically.

When a loan goes live, Confiance mints a score passport: a non-transferable NFT in the borrower’s wallet recording score, level, and status, valid for 90 days. Another lender can verify it with one public call or straight from the ledger without seeing personal data. The ledger refuses to transfer it, and its record is content-addressed, so changing one byte breaks the match. On default, the passport is revoked immediately.

Compliance comes first: a sanctioned wallet is stopped before any credit pull. Pulls are soft, consent-based, and store only scores and reason codes. A borrower with no credit file is offered a small secured credit-builder loan instead of a flat no; repaying it raises their score next time. Lenders are protected at every layer: ledger-enforced first-loss cover of at least 10% of debt and a debt maximum, a cap of 20% of available funds per loan, the sanctions gate, the ban list, an overdue monitor that alerts within seconds, and a human approving every loan. Everything runs on XRPL Devnet with no mock ledger; every action is a transaction on the explorer.

Our customer is a loan broker that wants XRPL credit without building underwriting, compliance and monitoring from scratch. The broker earns the management fee (10% of interest), a 1% origination fee, and a small per-payment service fee; depositors earn the rest. Revenue is reporting only and never feeds into approval.

The demo runs on Devnet with test money and keys held by our server; in production each party signs in their own wallet (GemWallet, CrossMark, or Xaman). Borrower histories are seeded, since new Devnet wallets have none. Next: a real KYC provider issuing the credential, two-key approval via multi-signing, on-chain collateral for secured offers, and passports pinned on IPFS. The foundation is in place: native XRPL lending with a decision layer people can trust because they can check it.