Execution Layer
Agent inference runs inside AMD SEV-SNP enclaves with deterministic replay. Operators are scheduled against the Solana leader schedule so your instructions land at the earliest tipped slot — the same class of hardware Jito runs its block engines on.