What Is OpenClaw?

OpenClaw is an open-source AI agent orchestration platform. Think of it as the operating system for your AI workforce. It manages agent sessions, schedules tasks, connects to your tools, and keeps everything running — even when you're sleeping.

Unlike chatbots that wait for your input, OpenClaw agents are autonomous. They have schedules, memory, and the ability to take actions across your tools (email, Slack, CRM, GitHub, and more).

Prerequisites

Before starting, you'll need:

  • A server or VPS — Any Linux machine works (Ubuntu recommended). Even a small VPS is fine to start.
  • An Anthropic API key — OpenClaw uses Claude as its AI brain. Get a key from Anthropic's console.
  • Basic terminal comfort — You don't need to code, but you should be able to run commands in a terminal.

Step 1: Install OpenClaw

SSH into your server and run:

curl -fsSL https://get.openclaw.ai | bash

This installs OpenClaw and its dependencies. The installer will walk you through initial configuration including your API key setup.

Step 2: Configure Your Workspace

OpenClaw uses a workspace directory (typically ~/.openclaw/workspace/) that acts as your agent's brain. Key files:

  • AGENTS.md — Rules and conventions for your agent
  • SOUL.md — Your agent's personality and identity
  • BRAIN.md — Active goals and tasks
  • TOOLS.md — Tool-specific configuration notes

Start by editing SOUL.md to give your agent a name and personality. This isn't just flavor — it affects how the agent communicates and makes decisions.

Step 3: Add Skills (Tool Integrations)

Skills are how your agent connects to external tools. Common skills include:

  • Email — Read inbox, send messages, draft responses
  • Calendar — Check schedules, create events, manage availability
  • Slack/Discord/Telegram — Communicate in team channels
  • GitHub — Create PRs, review code, manage issues
  • CRM — Update contacts, score leads, log interactions

Each skill has a SKILL.md file that documents how to configure it. Install skills with:

openclaw skill install email
openclaw skill install calendar
openclaw skill install slack

Step 4: Build Your First Agent

The easiest first agent is a morning briefing agent. Here's what it does:

  1. Wakes up at 7 AM
  2. Scans your email for important messages
  3. Checks today's calendar
  4. Reviews CRM for deals needing attention
  5. Delivers a summary to Slack or Telegram

Set this up by adding a cron schedule in your OpenClaw config:

# In your OpenClaw config
cron:
  morning-briefing:
    schedule: "0 7 * * *"
    prompt: "Good morning. Check email, calendar, and CRM. Deliver a briefing summary."
    channel: telegram

For a detailed walkthrough, see our morning briefing agent tutorial.

Step 5: Deploy and Monitor

Start the OpenClaw gateway:

openclaw gateway start

Your agent is now live. Monitor it with:

openclaw gateway status
openclaw gateway logs

For the first few days, review every output your agent produces. Tweak the instructions in your workspace files based on what you see. Most agents hit their stride within a week.

What to Automate Next

Once your morning briefing is running smoothly, expand to:

  • Lead outreach — Agent drafts and sends personalized outreach
  • Email triage — Agent categorizes and drafts responses to incoming email
  • Code reviews — Agent reviews PRs and suggests improvements
  • Report generation — Weekly business metrics delivered automatically

Learn more about scaling your agent workforce in our guide on automating your business with AI.

Need help with setup?

Johann Sathianathen offers OpenClaw consulting to get you running in days. Or join the community for free support from other operators.

Book a Call → Join the Community

Troubleshooting Common Issues

  • Agent not responding — Check openclaw gateway status and verify your API key is valid
  • Skills not connecting — Ensure OAuth tokens are configured in the skill's config file
  • Agent making errors — Refine instructions in SOUL.md and BRAIN.md. Be more specific about edge cases.
  • High API costs — Use model routing to send simple tasks to cheaper models

Resources