What Moss does
Moss is a real-time semantic search runtime built for conversational AI, developed by InferEdge Inc. It provides a local-first search engine that embeds directly inside AI applications, giving agents and copilots fast access to relevant context during conversations without the latency of remote database queries.
Key capabilities
Moss delivers sub-10ms lookup latency with instant index updates and a local-first architecture that keeps data on-device by default, with optional cloud sync. It supports hybrid search combining semantic similarity and keyword matching, ships built-in embedding models (moss-minilm and moss-mediumlm) alongside custom embeddings, and is built in Rust and WebAssembly for performance. It deploys across browsers, edge devices, and the cloud through a unified API, with JavaScript/TypeScript and Python libraries.
Who it's for
Moss serves teams building AI agents, voice assistants, and copilots that need fast conversational context retrieval. Common use cases include voice agents and copilots requiring real-time grounding, knowledge base and help-center search, on-device and offline applications, and edge deployments where a lightweight runtime matters. It is backed by Y Combinator.