How can we
help you build?
Everything you need to know about setting up your own AI agent gateway with OpenClaw.
What is OpenClaw?
OpenClaw is a locally hosted agent gateway that connects your AI logic to messaging platforms like WhatsApp, Telegram, and Discord in real-time. It provides a unified API for managing sessions, routing commands, and executing tools, allowing you to build autonomous agents that live where you chat. Learn about its core architecture and how it prioritizes data privacy through self-hosting.
Can I run multiple bots/agents?
Yes, OpenClaw is built for multi-agent orchestration. You can run multiple isolated agents on a single gateway, ensuring that valid sessions, memory, and workspaces are kept separate. Use flexible routing bindings to direct messages from specific channels or accounts to the correct agent automatically.
What happens if one model fails?
Reliability is baked into the core with a two-stage failover system. If an authentication profile fails or is rate-limited, OpenClaw automatically rotates to the next available key. If all keys fail, it seamlessly falls back to alternative models you've configured, ensuring your agent never goes silent due to provider outages.
Why is OpenClaw not replying?
A silent bot is usually due to a missing permission or a misconfigured binding. This guide walks you through a 60-second diagnostic ladder using the CLI to check gateway status, probe channel connectivity, and inspect live logs. We'll help you pinpoint issues like missing message intents in Discord or unapproved pairings in WhatsApp.
How do I create a custom hook?
Hooks allow you to extend OpenClaw's functionality by running custom TypeScript code in response to events. Learn how to create a simple handler that triggers on commands like /new or /reset, or even during gateway startup. We provide a step-by-step guide to setting up the directory structure and writing your first automation script.