Blog

OpenClaw’s heartbeat system lets your agent proactively check on things and take action without you asking. Here’s how it works and how to use it effectively.

What Is a Heartbeat?

A heartbeat is a scheduled message sent to your agent on a regular interval. The agent reads HEARTBEAT.md for instructions and acts on anything that needs attention. If nothing needs doing, it replies HEARTBEAT_OK silently.

What to Put in HEARTBEAT.md

• Check email and flag anything urgent
• Review calendar and remind of upcoming events
• Monitor RSS feeds and summarize new posts
• Check if websites are responding
• Post scheduled content automatically
• Build Reddit karma with helpful comments

Setting Up Heartbeats

In OpenClaw config, enable the heartbeat scheduler and set your interval. The agent wakes up on schedule, reads HEARTBEAT.md, executes anything actionable, and goes back to sleep.

The Key Principle

Heartbeats turn OpenClaw from a reactive assistant (you ask, it answers) into a proactive one (it watches, notices, and acts). This is what separates OpenClaw from a simple chatbot. For 24/7 heartbeat operation, you need OpenClaw running continuously — a DigitalOcean VPS or always-on home server works perfectly.

Recommended on Amazon: Homelab Book | Linux Command Line Book

OpenClaw stores everything in files — memory, skills, config, and history. This makes backup trivially simple. Here’s how to never lose your OpenClaw data.

What OpenClaw Stores

• MEMORY.md — curated long-term memory
• memory/YYYY-MM-DD.md — daily session logs
• SOUL.md, USER.md, AGENTS.md — personality and config
• TOOLS.md — credentials and external tool config
• HEARTBEAT.md — scheduled tasks

All of this lives in your workspace directory (typically ~/.openclaw/workspace).

Git Backup (Recommended)

cd ~/.openclaw/workspace
git init
git remote add origin your-private-repo-url
git add .
git commit -m "backup"
git push

Add a daily cron job to auto-commit. Your entire OpenClaw state is versioned and recoverable.

Cloud Sync

Syncing the workspace folder to Dropbox, iCloud, or Google Drive provides automatic real-time backup. On macOS, move the workspace folder into the cloud sync directory and symlink it back.

VPS Snapshots

If running on a VPS, take regular snapshots. DigitalOcean makes this one click from the dashboard. A weekly automated snapshot costs pennies and gives full recovery options.

🛒 Recommended: Automation Business Book | Productivity Desk Mat

Running OpenClaw in the cloud means your AI agent is always available, doesn’t drain your laptop battery, and can be accessed from anywhere. Three cloud providers stand out in 2026: DigitalOcean, Vultr, and Hetzner. Here’s an honest comparison.

DigitalOcean — Best Overall

The go-to for most OpenClaw users. Clean dashboard, excellent documentation, strong community, and generous new user credits. The $6/month Basic Droplet (2GB RAM, 50GB SSD) handles OpenClaw comfortably.

New users get $200 in free credits — that’s over 2 years of hosting at the entry tier.

Get started with DigitalOcean →

Vultr — Best Global Coverage

32 data center locations — more than any other provider on this list. Vultr’s High Frequency instances offer better CPU performance per dollar than comparable DigitalOcean tiers. If you need a server in a specific region (Johannesburg, Osaka, Seoul), Vultr likely has it.

Try Vultr →

Hetzner — Best Value in Europe

Hetzner offers extraordinary value for European users. Their CAX11 ARM instance (4GB RAM, 2 vCPUs) costs around €3.79/month — roughly half the price of comparable US providers. Latency from North America is higher, but for European users it’s the clear price winner.

Quick Comparison

• Best for beginners: DigitalOcean (cleanest onboarding, best docs)
• Best value globally: Hetzner (Europe) or Vultr (worldwide)
• Best free trial: DigitalOcean ($200) or Vultr (up to $250 promotional)
• Best performance per dollar: Vultr High Frequency

Recommended Starting Configuration

For most users: Ubuntu 22.04 LTS, 2GB RAM, 50GB SSD, nearest datacenter region. This handles OpenClaw with room to spare. Upgrade to 4GB if you plan to run browser automation tasks or multiple simultaneous operations.

Start with DigitalOcean — $200 free →

🛒 Recommended: Automation Business Book | Productivity Desk Mat

Telegram is the most popular channel for OpenClaw — it’s fast, reliable, works on every device, and has a well-designed bot API. Here’s how to connect OpenClaw to Telegram from scratch.

Step 1: Create a Telegram Bot

1. Open Telegram and search for @BotFather
2. Send: /newbot
3. Choose a name for your bot (this is the display name)
4. Choose a username ending in “bot” (e.g., myopenclaw_bot)
5. BotFather sends you an API token — save this

Step 2: Configure OpenClaw

Run openclaw setup and when prompted for channel, select Telegram. Enter your bot token. OpenClaw handles the rest of the connection automatically.

Step 3: Start a Conversation

Find your bot in Telegram (search for its username), send /start, and begin chatting. Your OpenClaw agent is now live on Telegram.

Privacy Settings

By default, your Telegram bot responds to anyone who knows its username. To restrict access to only yourself:

1. Find your Telegram user ID (message @userinfobot)
2. In OpenClaw config, set allowedUsers to your ID
3. OpenClaw ignores messages from any other user

Useful Commands After Setup

• Just type naturally — OpenClaw understands conversational requests
• Ask it to remember things: “Remember that my standup is at 9am on weekdays”
• Set it tasks: “Check my email and summarize anything important”
• Use skills: “What’s the weather in New York tomorrow?”

Keeping It Running 24/7

OpenClaw on Telegram needs to stay running to receive messages. For always-on availability, a DigitalOcean VPS is the most reliable option — new users get $200 free credit. Your bot stays online even when your laptop is closed.

🛒 Recommended: Automation Business Book | Productivity Desk Mat

The Mac Mini is one of the best machines for running OpenClaw. It’s quiet, power-efficient, runs macOS natively, and has enough power to run local AI models alongside OpenClaw if you want. Here’s the complete setup from scratch.

What You’ll Need

• Mac Mini (M2 or M4 recommended — check current price)
• macOS 13 Ventura or later
• A messaging channel (Telegram bot is easiest)
• About 30 minutes

Step 1: Install Homebrew

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

Step 2: Install Node.js

brew install node

Step 3: Install OpenClaw

npm install -g openclaw

Step 4: Run Setup

openclaw setup

Follow the prompts to connect your Telegram bot (or other channel) and configure your AI provider.

Step 5: Start OpenClaw

openclaw start

Step 6: Run as a Background Service

To keep OpenClaw running 24/7 and restart automatically after reboots:

openclaw service install
openclaw service start

Power Settings

Go to System Settings > Energy > and disable “Put hard disks to sleep when possible” and set the Mac to never sleep. This ensures OpenClaw stays responsive at all times. The Mac Mini uses about 15W at idle — roughly $1-2/month in electricity.

Remote Access

Enable Screen Sharing (System Settings > General > Sharing) so you can access your Mac Mini remotely if needed. For secure remote access from outside your home network, Tailscale is the cleanest solution.

Running OpenClaw with Ollama means your AI agent operates entirely on your own hardware, with no data leaving your machine. No API costs, no privacy concerns, no dependence on external services. Here’s how to set it up.

What is Ollama?

Ollama is a tool for running large language models locally. It handles the technical complexity of loading and running models like Llama 3, Mistral, and Gemma on your hardware. For OpenClaw, it becomes the AI backend instead of cloud services like Claude or GPT.

Hardware Requirements

Local AI models need RAM — specifically:

• Llama 3.2 3B: ~4GB RAM (fast, capable for most tasks)
• Mistral 7B: ~8GB RAM (excellent quality/speed balance)
• Llama 3.1 8B: ~8GB RAM (strong reasoning)
• Llama 3.1 70B: ~40GB+ (best quality, needs high-end hardware)

A Mac Mini M4 with 16GB RAM handles Mistral 7B comfortably alongside OpenClaw. Mac Mini M4 on Amazon.

Installation

Step 1: Install Ollama

curl -fsSL https://ollama.ai/install.sh | sh

Step 2: Pull a Model

ollama pull mistral
ollama pull llama3.2

Step 3: Configure OpenClaw to Use Ollama

In your OpenClaw config, set the AI provider to Ollama and point it to the local endpoint (default: http://localhost:11434). OpenClaw handles the rest.

Performance Expectations

Local models are slower than cloud APIs on most hardware. Expect 5-20 seconds for responses on an 8B model, versus under 3 seconds with Claude API. For non-time-sensitive tasks (email summaries, research, content drafts), this is perfectly acceptable. For real-time conversation, cloud APIs are faster.

The Hybrid Approach

Many users run Ollama for routine tasks (cheaper, private) and fall back to Claude or GPT-4 for complex reasoning. OpenClaw supports this configuration. It’s the most cost-effective approach for heavy users.

Both OpenClaw and Home Assistant run on your own hardware and give you more control than cloud-based alternatives. But they solve very different problems. Here’s a clear breakdown of what each does and which one you should choose.

What Home Assistant Does

Home Assistant is a home automation platform. It integrates with smart devices — lights, thermostats, locks, sensors — and lets you create automations between them. It excels at physical-world automation: “turn off all lights when I leave” or “alert me if the door opens while I’m asleep.”

What OpenClaw Does

OpenClaw is an AI agent runtime. It runs a large language model (Claude, GPT-4, Gemini) that can use tools, remember context across sessions, and act autonomously on your behalf. It excels at knowledge work: managing communications, writing content, answering complex questions, and coordinating multi-step tasks.

The Key Differences

• Interface: Home Assistant is primarily visual (dashboard, automations editor). OpenClaw is primarily conversational (Telegram, Discord, Signal).
• Integration focus: Home Assistant integrates with physical devices. OpenClaw integrates with digital services and APIs.
• AI capability: Home Assistant has limited AI features. OpenClaw is built around AI as its core capability.
• Learning curve: Both require setup, but Home Assistant’s ecosystem is more complex for advanced use.

Can You Run Both?

Yes — and many people do. Home Assistant handles the physical home (lights, climate, security). OpenClaw handles the digital life (communications, tasks, content). They complement each other rather than competing. A Raspberry Pi 5 or Mac Mini can run both simultaneously. Raspberry Pi 5 for home server.

Which Should You Start With?

Start with OpenClaw if you want an AI assistant that handles communication and knowledge work. Start with Home Assistant if your primary goal is smart home automation. For always-on deployment of either, a DigitalOcean VPS ($200 free credit for new users) or a dedicated home server works well.

OpenClaw runs on Windows 11. Here is everything you need to get started.

Step 1: Install Node.js

Download the LTS version from nodejs.org and install with default settings. Verify: open Command Prompt and type node --version.

Step 2: Install OpenClaw

Open Command Prompt as Administrator:

npm install -g openclaw

Step 3: Setup and Start

openclaw setup
openclaw start

Follow the prompts to connect your Telegram or other messaging channel.

Run on Startup

Press Win+R, type shell:startup, create a file called openclaw.bat with the line openclaw start. OpenClaw will start with Windows automatically.

Windows vs VPS

Running on Windows works well if your PC stays on. For always-on availability without your PC running, a DigitalOcean VPS at $6/month (with $200 free credit) is the better choice.

🛒 Recommended: Automation Business Book | Productivity Desk Mat

Both DigitalOcean and Vultr run OpenClaw excellently. Here is an honest comparison.

Quick Verdict

• Best for beginners: DigitalOcean — cleaner UI, better docs
• Best value / global coverage: Vultr — 32 locations, competitive pricing

Pricing

Both start at ~$6/month for 2GB RAM. Vultr’s High Frequency instances offer better CPU per dollar.

Ease of Use

DigitalOcean wins here. The dashboard is the cleanest in the industry. Perfect for first-time VPS users.

Welcome Credits

DigitalOcean: $200 free credit. Vultr: up to $250 during promotions.

Our Pick

Start with DigitalOcean if you are new to VPS hosting. The documentation and $200 credit make it the easiest entry point.

Get started with DigitalOcean ($200 free) →

For maximum global coverage: Try Vultr →

🛒 Recommended: Automation Business Book | Productivity Desk Mat

DigitalOcean is the fastest way to get OpenClaw running in the cloud. New users get $200 in free credits — enough for 2+ years at the entry tier.

Step 1: Create Your Account

Sign up at DigitalOcean and claim your $200 credit.

Step 2: Create a Droplet

Click Create → Droplets. Choose Ubuntu 22.04 LTS, Basic plan at $6/month (2GB RAM), and your nearest datacenter.

Step 3: Connect via SSH

ssh root@YOUR_IP

Step 4: Install Node.js

curl -fsSL https://deb.nodesource.com/setup_20.x | sudo -E bash -
sudo apt-get install -y nodejs

Step 5: Install OpenClaw

npm install -g openclaw
openclaw setup
openclaw start

Step 6: Keep It Running 24/7

npm install -g pm2
pm2 start openclaw --name openclaw
pm2 startup && pm2 save

Your OpenClaw is now live 24/7. Get your $200 DigitalOcean credit →

🛒 Recommended: Automation Business Book | Productivity Desk Mat