The Bloomberg Terminal for Machines.
Traditional oracles are price-feed centric and slow. Agents need non-deterministic data in sub-second intervals.
Existing oracles like Pyth and Chainlink are built for price feeds. They're not designed for the questions AI agents actually ask: "Did Trump just tweet?" "What's the result of this API call?" "Is this inference output valid?"
Vox is designed to be the fastest data intake for agents. An Optimistic Oracle architecture that posts data on-chain immediately, then verifies through a short dispute window.
"Post first. Verify later."
The core problem: Solana's block time is 400ms. Legacy oracles can't keep up. Agents end up making decisions on stale data. Vox solves this with Optimistic Posting—data is available immediately, and if no dispute is raised within the verification window, it becomes canonical.
The_Problem
Agents need to know: "Did X happen?" not just "What's the price of Y?"
Legacy oracles update too slowly. By the time data is "confirmed," it's already stale for high-frequency operations.
This is the Stale Data Risk.
The_Solution
Optimistic Posting: Data is reflected on-chain immediately.
Dispute Window: A short verification period where watchers can challenge incorrect data.
Slashing: Publishers back their data with stake. Incorrect data = stake forfeited.
.gif&w=3840&q=75)