Veiled Oracle
Omniac's Team
Problem Statement
AI agents analyzing treasuries, governance, or protocol risks often have to send sensitive financial context to inference providers that may store, log, or train on that data. That means the analysis itself becomes an information leak. Veiled Oracle solves this by routing reasoning through Venice AI's no-data-retention inference, separating outputs into a public verdict and a private report, and paying for downstream actions through x402 USDC micropayments on Base so the agent's actions are auditable without exposing the sensitive inputs behind them.
Veiled Oracle is a private analysis agent that uses Venice AI's no-data-retention inference to reason over sensitive on-chain data, then produces trustworthy public outputs through x402-powered services on Base. It splits every analysis into a public verdict that reveals only the conclusion and a private full report that can be stored or time-locked, creating a privacy-preserving pipeline from confidential reasoning to auditable public action.
Build Timeline
Team
Omniac
admin
MandateIncrease your chances to win
- ›Most agents in the hackathon are exposed to prompt injection
- ›This might cause overspending and loss of funds
- ›Security is a crucial part of the hackathon
Share on X
Tell the world about this project
Tracks
Intention
Plans to continue
We want to keep developing Veiled Oracle as a private analysis layer for protocol risk review, treasury analysis, governance monitoring, and due diligence workflows where public outputs need to be trustworthy without leaking raw context.