TrustedRouter has raised $6 million in seed funding led by Sierra Ventures as it enters the expanding market for artificial intelligence model routing. The investment comes as businesses seek greater control over AI costs, providers, and the handling of sensitive information. The startup is emphasizing privacy and verifiable security to distinguish itself from established competitors.
Funding and Company Background
Founded by chief executive Joseph Perla, Miami-based TrustedRouter gives developers access to AI models through a common application programming interface. Customers can choose providers according to price, performance, availability, region, and data-handling requirements. The approach simplifies applications that would otherwise need separate integrations with different model vendors.
TrustedRouter previously announced $1.25 million in seed backing in August from investors including Sam Lessin, Bill Tai, Linda Avey, and George Xing. It has not publicly clarified whether the newly reported $6 million includes that earlier funding or represents separate financing. The investment reflects interest in infrastructure for managing increasingly diverse AI models.
Privacy as a Competitive Strategy
TrustedRouter promotes an open-source gateway and confidential-computing architecture designed to protect sensitive customer requests. Its gateway uses hardware attestation so users can verify the software environment handling their prompts, according to the company. Ordinary inference requests are not retained in prompt or response logs, although operational metadata is collected.
On certain supported routes, protections extend beyond TrustedRouter's gateway to the infrastructure running the AI model. These configurations combine encryption and verified confidential computing to prevent the intermediary and hosting provider from inspecting requests or outputs. Such guarantees, however, do not apply automatically to every model or provider on the platform.
Customers can set routing preferences for data retention, location, and confidential execution. A provider's commitment not to retain prompts differs from technically verified protection while those prompts are being processed. TrustedRouter says that when no route satisfies a customer's required privacy level, the request is rejected rather than redirected to a less secure provider.
Competition in AI Model Access
TrustedRouter faces OpenRouter, an established platform connecting developers to hundreds of AI models through one interface. In August, Stripe announced an agreement to acquire OpenRouter, highlighting the commercial importance of tools that optimize AI usage and spending. The transaction also signals intensifying competition to control the infrastructure connecting applications and model providers.
According to TrustedRouter, models from DeepSeek, Zhipu, and Moonshot are among the popular choices on its platform. New open-weight releases from companies including Mistral and Reflection AI could give customers further options. However, developers must assess each model's quality, reliability, and privacy characteristics before shifting business-critical workloads.
Plans for an AI Token Marketplace
Perla has proposed developing a competitive marketplace for AI inference capacity that he calls the New York Token Exchange. In that vision, data centers hosting open-weight models would compete to process individual requests based on pricing and other requirements. He has also discussed eventual financial instruments tied to AI tokens, including price indices and hedging products.
The proposed exchange is a long-term ambition, not an operating financial marketplace. For customers today, TrustedRouter's more immediate offering is the ability to compare providers and redirect workloads without rebuilding each integration. The savings and security benefits will depend on the models selected, available infrastructure, and the specific demands of each application.
TrustedRouter's $6 million funding highlights investor interest in flexible AI infrastructure with stronger data protections. Its focus on verifiable privacy provides a potential competitive advantage, although full confidential-inference protections remain limited to supported routes. Sustained growth will depend on delivering dependable model access, competitive economics, and effective security as routing services become more widely adopted.