The problem
0G's compute network is powerful, but onboarding breaks the moment an app forces users to bridge assets and juggle a native token just to run a model. That's friction that kills adoption before it starts.
What 0friction does
It abstracts the whole mess into a developer primitive. The user never touches 0G tokens or a bridge — they sign a gasless authorization on the chain they're already on, pay in USDC, and get their inference result back. A solver handles execution on 0G and settlement on the home chain.
What's in the box
@0friction/sdk— TypeScript SDK for any runtime (published on npm)- Solver backend — 0G execution + USDC settlement
- IntentRegistry contract — deployed on 0G Galileo
- Demo frontend + CLI agent — a working example end to end
Developer experience
Get a quote, build an intent, the app layer signs intent + permit with viem/ethers/wagmi, submit. The SDK is designed so an agent can request AI compute the same way it would call any other API — that's the whole point.