The Lending Network for Real World Assets
Agama is a lending network for real world assets, built entirely in Arbitrum Stylus (Rust). Lenders deposit USDC and receive agUSD one to one, then stake it into sagUSD to earn real yield that compounds straight into the share price, with nothing to claim. Borrowers post tokenized private credit vault shares from Qiro Finance and Tenka, spanning ABF senior and mezzanine tranches, payment financing and institutional credit, and borrow USDC against them up to each vault's LTV.
Interest follows an Aave style kinked utilization curve, and a NAV oracle prices every RWA vault on chain, accruing each second, so the protocol stays solvent without relying on a keeper. A credit event, when a vault's NAV drops, is what pushes position into liquidation.
Liquidations are where Agama is different. In RWA backed lending a late liquidation is bad debt, so the fastest liquidation has to win. A keeper watches every position and the instant a health factor breaks one, fires the liquidation through the Arbitrum Timeboost express lane, sequenced a full block ahead of every competing bot. This is proven end to end on Arbitrum Sepolia, with the real liquidate call landing with timeboosted set to true, reproducibly.
During the buildathon we locked in partnerships across the RWA and infra stack: bond.credit for a 24/7 AI Watchtower liquidation agent, and Qiro Finance and Fatra Labs for real, yield bearing private credit collateral. On traction, Agama won the Rayls Hackathon at ETHCC Cannes 2026, has over $4M in private credit from signed issuers ready to deploy into our vaults and over $5M in LP capital in discussion, and is a Rayls Launch Partner alongside Enzyme, LayerZero, Lagoon Finance and Liqvid. On top of that we shipped the full protocol on Arbitrum: agUSD, sagUSD, NavOracle, the RWA vaults and a LendingPool, all written in Stylus and reproducibly verified on Arbitrum Sepolia, with the full lifecycle proven on chain and real Timeboost liquidations landing with timeboosted: true.
Not currently raising, bootstrapped
