Bring your framework.
We operate & govern it.
Build agents in LangChain, LangGraph, CrewAI, AutoGen, LlamaIndex, or Haystack — then run them on Refusion for scoped identity, smart model routing, spend caps, and a single audit stream. One control plane for every agent ecosystem.
Model routing
Smart per-task routing with automatic failover — same router for every framework.
Spend caps
Hard per-agent and swarm-wide caps, enforced before any run starts.
Audit & govern
Every run, tool, and cost is logged to one auditable stream.
Orchestration
Chain framework agents into supervisor swarms with shared state.
Supported frameworks
operate & govern on RefusionLangChain
Chains & tool-callingCompose tools, memory, and retrieval into runnable chains.
Refusion wraps each chain run with smart routing, a spend cap, and a full audit trail.
LangGraph
Stateful multi-agent graphsModel complex agent state machines with cycles and branches.
Operate LangGraph runs behind Refusion governance — every node routed and metered.
CrewAI
Role-based crewsDefine role-driven crews that collaborate toward a goal.
Give each crew member a scoped Refusion key, spend cap, and task route.
AutoGen
Conversational multi-agentMulti-agent conversations with code execution.
Meter every AutoGen exchange and enforce caps before any tool fires.
LlamaIndex
Data & retrieval (RAG)Index, retrieve, and query private data for grounded LLMs.
Govern retrieval-augmented runs with Refusion routing and audit.
Haystack
Search & QA pipelinesBuild production search and QA pipelines.
Wrap Haystack pipelines in a governed, routed, audited control plane.
Operate & govern any framework — included on Scale & Enterprise.
Route, cap, meter, and audit agents built in any framework above. Provisioned through one console, governed by one policy.