Every AI-powered product faces the same trade-off: which model should handle each request? TrustedRouter, a startup that launched publicly six weeks ago, just closed a $6 million seed round led by Sierra Ventures to answer that question automatically, reports Semafor.
The platform routes developer and business requests across AI models based on factors like cost and privacy. For some open-weight models, it offers end-to-end encryption and verifiable confidential computing, designed so that neither the company nor the data center can read customer prompts or outputs. The pitch is simple: stop locking every workload to one provider, and let a routing layer pick the best model for each job.
TrustedRouter also plans a marketplace called the New York Token Exchange, where models and hosting providers would compete to process requests, pricing AI tokens like commodities. That is an ambitious vision, but it signals where the market is heading: infrastructure that treats model capacity as a tradable utility.
The funding lands as enterprises scale AI across support and sales workflows. Teams running tools like Intercom Fin or Zendesk AI increasingly depend on multi-model stacks, and a neutral routing layer could cut costs and vendor lock-in. If you are evaluating tools for your own stack, our AI Tool Buyer’s Guide breaks down what to check first.
Source: AI Brief (via Semafor)
