Summary
Moss builds a real-time semantic search runtime for conversational and voice AI that delivers sub-10 ms retrieval. Founded by Sri Raghu Malireddi and Harsha Nalluru, who bring experience from Grammarly and Microsoft, they solve the retrieval lag that makes AI interactions feel unnatural. Built in Rust and WebAssembly, Moss runs natively across browsers, mobile devices, and servers, keeping search close to where agents run with zero network hops. They're powering production pilots across voice AI and developer platforms, achieving 70-90% token savings compared to traditional pipelines.