TrustedRouter
TrustedRouter provides a hardware-attested, zero-log AI gateway that routes API calls to 100+ LLM providers with cryptographically verifiable privacy. The platform runs inside GCP Confidential Space, allowing users to verify the running image against open-source code before any prompt is processed. Migration requires only changing the base URL to https://api.trustedrouter.com/v1 while keeping existing OpenAI-compatible SDK integrations.
- Artificial Intelligence
- Cybersecurity
- Developer Tools
- Software Only
Funding
Founders
Founder details are not available yet.
Product
Problem
Companies using hosted LLM APIs face a fundamental trust gap: providers claim privacy but offer no verifiable way to confirm that prompts and outputs are not logged, retained, or exposed. This forces organizations with sensitive data—legal documents, medical records, proprietary code—to either accept unverifiable privacy claims or forego the benefits of frontier models entirely.
Solution
TrustedRouter provides an end-to-end encrypted AI gateway that routes requests to 100+ LLM providers through a hardware-attested, zero-log infrastructure. The gateway runs inside GCP Confidential Space, where hardware-backed attestation cryptographically signs the running image so users can verify that only the open-source code handles their prompts. The platform offers routing aliases such as trustedrouter/zdr for zero-data-retention routes prospected for legal and sensitive work, trustedrouter/e2e for confidential compute with provider-side E2EE, and trustedrouter/eu for Europe-focused routing. Migration requires only changing the base URL to https://api.trustedrouter.com/v1 while keeping the OpenAI SDK and existing application architecture. The gateway is fail-closed: if it cannot prove its own integrity, traffic stops, with a synthetic monitor probing attestation every minute and paging on the first miss.
Target Audience
Primary users are engineering teams and organizations handling sensitive data—legal, financial, healthcare, and enterprise development—who need verifiable privacy guarantees when routing to LLM providers, as well as AI agents and applications already using OpenAI-compatible SDKs.
Features
- Hardware-attested gateway running in GCP Confidential Space with signed image digests verifiable against public source commits
- Zero prompt or output logging—the binary never opens a write handle to a prompt path, with the schema for metadata-only billing records in open-source code
- Routing aliases including trustedrouter/auto for healthy-provider rollover, trustedrouter/zdr for zero-retention routes, trustedrouter/e2e for confidential compute, and trustedrouter/eu for European providers
- 603 public models across 114 providers with 1,808 configured routes, including 431 ZDR routes and 56 provider E2EE routes
- OpenAI-compatible and Anthropic-compatible endpoints supporting one base_url migration from existing SDK integrations
- Per-minute synthetic monitoring that probes attestation integrity and pages on first miss, with fail-closed traffic behavior
- Provider failover and regional routing to prevent upstream outages from becoming product outages
- Signed inference receipts providing offline-verifiable proof of every response, including exact bytes and hardware-attested signing key
- Provider privacy directory publishing each provider's known retention and ZDR posture per model route
- Fail-closed privacy filters via provider.min_privacy parameter supporting "zdr" and "confidential" (verified compute + E2EE + ZDR) hard minimums