semantic-router

What problem does it solve?

Route user queries and agent requests quickly and deterministically using semantic vector matching to avoid slow LLM-only routing, reduce latency, and improve precision for intent classification and tool invocation.

Core Features & Use Cases

• High-speed intent classification: Match queries to predefined routes using embedding similarity instead of expensive generation.
• Dynamic function calling: Extract parameters and generate function-call schemas for executing tasks programmatically.
• Flexible index and encoder support: Work with local encoders or API encoders and backends like Local, Pinecone, Qdrant, or Postgres for production.
• Use Case: Build a chatbot routing layer that identifies user intent, blocks sensitive topics, and triggers appropriate backend functions with extracted parameters.

Quick Start

Use the semantic-router to classify an input query and return the best-matching route name plus any function call arguments.