1RPC.ai LLM Relay
Overview
Learn how the 1RPC.ai LLM relay was designed to protect AI model requests.
What is 1RPC.ai?
1RPC.ai explores how a Trusted Execution Environment (TEE) can protect requests routed to verified AI models. A TEE provides hardware-based isolation designed to preserve the confidentiality and integrity of the computation that runs inside it.
The relay design is based on three principles:
- Verify that upstream model endpoints are operated by their stated providers.
- Use only the minimum metadata required for routing, then wipe it after relaying.
- Sign requests from a TEE relay so its execution guarantees can be verified.
This section documents the privacy principles and attested relay design behind 1RPC.ai. Visit the 1RPC.ai landing page for the current project overview.