TrustedRouter

Product Introduction

1. Definition: TrustedRouter is an open-source, privacy-first AI gateway and unified API platform. Technically, it functions as an abstraction layer and routing engine that sits between a developer's application and over 600 large language models (LLMs) from more than 90 providers.
2. Core Value Proposition: It exists to solve the fragmentation, privacy, and reliability challenges in the AI inference market. Its primary value is delivering verifiable privacy with proof, provider-agnostic routing, and cost optimization through a single, OpenAI-compatible API endpoint.

Main Features

1. Attested Gateway with Zero-Content Logging: The core of TrustedRouter's privacy promise. The production gateway runs in a Trusted Execution Environment (TEE) and is cryptographically attested. This process generates public evidence linking the running binary to a specific, auditable open-source commit. Critically, the prompt path is engineered to exclude prompt and output content from all logging and control plane systems, a claim users can independently verify.
2. Explicit Privacy & Geographic Routing Aliases: Instead of just model names, developers can route requests based on privacy and compliance requirements using predefined aliases. trustedrouter/zdr routes only to providers with a zero-data-retention policy. trustedrouter/e2e selects providers offering confidential computing with end-to-end encryption. trustedrouter/eu ensures routing through European infrastructure for GDPR and data sovereignty.
3. Multi-Cloud Provider Failover & Health Monitoring: The infrastructure spans GCP, AWS, and Azure, ensuring no single cloud dependency. The system continuously measures provider and model latency and availability. The trustedrouter/auto alias implements intelligent, healthy-provider rollover, preventing a single upstream provider outage from causing downstream application failures.
4. OpenAI-Compatible Unified Interface: The API is fully compatible with the OpenAI SDK schema. Migration for most applications requires changing only the base_url parameter to https://api.trustedrouter.com/v1, allowing developers to instantly access hundreds of models without refactoring code.
5. Bring Your Own Key (BYOK) and Transparent Pricing: Users can bring their existing API keys from supported providers (e.g., Anthropic, OpenAI, Google) to maintain direct billing relationships or enterprise rates while leveraging TrustedRouter's routing and privacy layer. The platform operates on a pay-as-you-go credit system with no subscription fees, and routing costs are visible before API calls are made.

Problems Solved

1. Pain Point: Vendor Lock-in and Reliability Risk. Relying on a single AI provider's API creates business risk due to potential outages, rate limit changes, or price hikes.
2. Pain Point: Unverifiable Privacy Claims. Many AI gateways operate as "trust-me" black boxes, leaving enterprises in legal, healthcare, or finance unable to prove that sensitive prompts and outputs are not logged or misused.
3. Pain Point: API Fragmentation and Integration Complexity. Managing multiple API keys, different SDKs, and varying endpoints for models from OpenAI, Anthropic, Google, and open-weight leaders like DeepSeek and Qwen creates significant developer overhead.
4. Target Audience: Enterprise Development Teams in regulated industries (legal, finance, healthcare), AI Startup Engineers needing reliable, cost-effective inference, Privacy-Conscious Developers, and Platform Builders integrating multi-model capabilities.
5. Use Cases: Sensitive Document Analysis: A law firm (like the featured Robot, Robot & Human) processing privileged legal documents using AI for extraction and reasoning without data retention risk. Mission-Critical Chatbots: A customer service agent that must remain online, automatically failing over to a backup model if the primary provider is slow or down. Global Compliance: A European company ensuring all AI processing for its users occurs within EU borders using the trustedrouter/eu route.

Unique Advantages

1. Differentiation vs. OpenRouter/Other Gateways: While others aggregate models, TrustedRouter's core differentiator is its verifiable, attestation-based trust model and fail-closed architecture. If the gateway attestation fails, it does not silently route traffic; it fails closed. It also offers more explicit, policy-based routing (zdr, e2e, eu) compared to general model aggregation.
2. Key Innovation: The "Trusted Execution Environment (TEE) Attestation for AI Routing" is the foundational innovation. It moves the value proposition from mere convenience to provable security. The integration of this with a multi-cloud, health-aware routing engine and a 100% open-source codebase creates a transparent and resilient system that competitors do not fully replicate.

Frequently Asked Questions (FAQ)

1. How does TrustedRouter's privacy compare to OpenRouter? TrustedRouter provides verifiable privacy through an attested gateway running in a Trusted Execution Environment with zero content logging, whereas OpenRouter's privacy model is based on policy and trust. TrustedRouter offers public evidence linking its production code to open-source commits for independent audit.
2. Can I use my existing OpenAI or Anthropic API key with TrustedRouter? Yes, through the Bring Your Own Key (BYOK) feature. This allows you to maintain your direct contracts and rate limits with providers like OpenAI or Anthropic while using TrustedRouter for routing, failover, and its privacy-enhancing gateway layer.
3. What happens if the TrustedRouter attested gateway fails or cannot be verified? The system is designed to fail closed. In the event of an attestation failure or inability to verify the trusted environment, the gateway will not process or route prompt traffic, preventing any possibility of requests flowing through an unverified, potentially insecure path.
4. Is TrustedRouter truly an OpenAI API drop-in replacement? For the vast majority of chat completions and Responses API calls, yes. You change the base_url to https://api.trustedrouter.com/v1 and use your TrustedRouter API key. Advanced routing features are accessed through the extra_body parameter or by using specific model aliases like trustedrouter/zdr.
5. How does provider failover work with the trustedrouter/auto model? The trustedrouter/auto alias continuously monitors the health, latency, and availability of all integrated providers and models. It intelligently routes your requests to the healthiest available endpoint that matches your criteria, providing automatic resilience against provider outages or performance degradation.