Introduction
A unified LLM gateway + knowledge (RAG) platform, run from one admin panel.
The platform is a gateway that unifies many LLM providers behind one API, a RAG system that turns documents into searchable knowledge and MCP servers, and a single admin panel that runs all of it.
#Three pillars
One OpenAI-compatible API. Routes requests to the right model via nodes, model mappings, model groups and advanced flow routing.
Turns documents and datasources into vector knowledge and generates searchable MCP servers.
The single interface that manages the entire gateway and knowledge stack with live observability.
#The journey of a request
A client (Cursor, Claude, your own app) sends an OpenAI-format request to the gateway. The gateway resolves the model — either a direct model or a model group; if the group uses the "flow" strategy, the request travels through the router graph detailed below to the right member and node.