Skip to main content

Overview

The mismatch Doga solves

Crypto-asset wealth grows fastest in places where local currencies weaken fastest. Kenya, Nigeria, and Ghana rank in the global top 20 for per-capita crypto adoption. The same holders pay rent, tuition, and medical bills in KES, NGN, and GHS.

Existing crypto-lending protocols (Aave, Compound, MakerDAO) accept the crypto but lend in USD-stablecoins. The borrower is left with USD-pegged debt and local-currency expenses — every move in USD/KES quietly reshapes their loan.

Doga keeps the borrower's liability in the currency they actually owe in.

How the loan flows

Two halves, with a clean trust boundary:

  • On-chain — collateral custody, fiat-denominated debt accounting, liquidation. Permissionless.
  • Off-chain — PSP integration, KYC, fiat reconciliation. Run by a single OPERATOR_ROLE whose only on-chain capability is recording fiat movements after PSP receipts.

Carry trade economics

Doga's revenue is the spread between two interest rates.

note

Cost of capital: USDC liquidity providers (the protocol borrows USDC at ~5-8 % APR to settle liquidations and bridge fiat). Lending rate: local-currency APR per market — currently 20 % KES, 30 % NGN, 25 % GHS. Spread: funds FX risk + operational overhead + protocol revenue.

The exact spread is set per currency in the FiatCurrencyRegistry. Higher FX-volatility currencies (NGN) carry higher APR.

Why three layers of code

LayerWhatWhy
Smart contractsCustody + accounting + liquidation logicTrustless. The only thing that decides where collateral goes.
REST APIRead chain state, return unsigned calldataLets frontends and backend services talk to Doga without re-implementing ABI handling. Holds zero keys.
MCP serverWraps the API for AI agentsLets Claude Desktop / Code etc. operate the protocol via 75 typed tools. Also keyless.

Each layer is independent. You can build directly on the contracts; the API and MCP are conveniences, not gates.

What's next

  • For a deeper walk-through of the system map: Architecture
  • For who-can-do-what-on-chain: Roles
  • For every term defined exactly once: Glossary