> ## Documentation Index
> Fetch the complete documentation index at: https://docs.agentrpc.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Introduction

> Connect AI agents to any function with AgentRPC

AgentRPC is a universal RPC layer for AI agents. It allows you to connect to any function, in any language, across network boundaries.

## Why AgentRPC?

You might have services deployed in a private VPC, on Kubernetes, even across multiple clouds. AgentRPC gives you the tools to wrap your functions in a universal RPC interface, connecting them to a hosted RPC server that can be accessed from open standards like:

* Model Context Protocol (MCP)
* SDKs with OpenAI-compatible tool definitions (e.g. OpenAI, [Anthropic](https://docs.anthropic.com/en/api/openai-sdk), LiteLLM, OpenRouter, etc.)

<img src="https://mintcdn.com/agentrpc/R37bnrP2OCO55b1T/images/deployment.png?fit=max&auto=format&n=R37bnrP2OCO55b1T&q=85&s=97a9656ccf71c8132dbc6a7095a7d9c8" alt="AgentRPC Deployment" width="1453" height="1669" data-path="images/deployment.png" />

## Key Features

<CardGroup cols={2}>
  <Card title="Multi-language Support" icon="code">
    Connect Agents to tools in multiple languages including TypeScript, Go, and Python (.NET coming soon).
  </Card>

  <Card title="Private Network Support" icon="lock">
    Register functions and APIs even in private VPCs with no open ports required.
  </Card>

  <Card title="Long-running Functions" icon="clock">
    Long polling SDKs allow you to call functions beyond HTTP timeout limits.
  </Card>

  <Card title="Full Observability" icon="chart-line">
    Tracing, metrics, and events with our hosted platform for complete visibility.
  </Card>

  <Card title="Automatic Failover" icon="shield">
    Platform keeps track of function health, performs automatic failover, and retries.
  </Card>

  <Card title="Framework Compatibility" icon="puzzle-piece">
    Out-of-the-box support for MCP and OpenAI SDK compatible agents.
  </Card>
</CardGroup>