What happens when you send a message: you write "Summarize my week" on Telegram → the message reaches OpenClaw on your server → OpenClaw asks the AI model through Venice.ai → the answer comes back to your chat. All in a few seconds.

• Always on — the server never sleeps, so your agent is reachable 24/7, even when your laptop is off.
• Low cost — from about $6/month for the server, plus a few dollars of AI usage.
• No hardware to buy — no need for a powerful computer at home.
• You're in control — you choose the AI model and can switch any time.
• Room to grow — start with one simple assistant, add more agents later.

These are my referral links: you get the discount or bonus, and I may receive a small reward at no extra cost to you. It helps me keep making free guides like this one.

Hostinger sells a ready-made product called Managed OpenClaw (hostinger.com/managed-openclaw). OpenClaw comes pre-installed — you pay, it switches on, and you connect your chat app from a visual dashboard.

Pick a plan size. For one to three AI agents, KVM 2 is the sweet spot.

Your server needs an operating system, just like your laptop. Hostinger will offer several. Choose Ubuntu 24.04 LTS (or 22.04 LTS).

• Ubuntu ✅ — the most popular choice for servers, with the most tutorials and help online.
• Debian — also fine, very similar to Ubuntu.
• Windows ❌ — available, but not recommended: it costs more, uses more memory, and all OpenClaw instructions are written for Linux.

SSH is like a secure phone call to your server: you type commands on your computer, and they run on the server. Choose one of these three ways:

ssh root@YOUR_SERVER_IP
# Example: ssh root@185.199.110.153
# It will ask for your password → paste the one from Hostinger.
# You WON'T see the characters while typing the password. That's normal!
# First time only: type "yes" if asked to trust the server.

whoami      # should print: root
free -h     # how much memory the server has
df -h       # how much disk space is free

Docker packs a program together with everything it needs into a sealed box called a container.

• Without Docker: you install many separate pieces by hand and hope they work together.
• With Docker: one command, and OpenClaw starts with everything included.
• Updating is just downloading the newer box.
• If something breaks, you throw the box away and start a fresh one — the rest of your server stays untouched.

1. In hPanel, open your VPS and click Docker Manager in the side menu.
2. Open the Catalog and search for OpenClaw.
3. Click Deploy — the settings are pre-filled.
4. Wait 1–2 minutes.

2a. Install Docker (one-time):

curl -fsSL https://get.docker.com | sh
docker --version   # should print something like: Docker version 27.x

2b. Start OpenClaw:

docker run -d \
  --name openclaw \
  --restart unless-stopped \
  -p 18789:18789 \
  -v openclaw-data:/root/.openclaw \
  openclaw/openclaw:latest

2c. Check it's running:

docker ps
# Look for "openclaw" with STATUS "Up ..." → all good.
# If it says "Exited", see what went wrong with:
docker logs openclaw

OpenClaw has a web dashboard. On your own computer, open any browser and go to:

http://YOUR_SERVER_IP:18789

It will ask for a gateway token (a password for the dashboard). Find it on the server:

docker logs openclaw 2>&1 | grep -i token

All of OpenClaw's settings live in one file. You'll fill it in over the next two steps. Here's what each part does:

{
  // 1) THE BRAIN — which AI service to use (Step 3)
  "providers": {
    "venice": {
      "apiKey": "vnc-...",                        // your secret Venice key
      "defaultModel": "venice/deepseek-v4-flash"  // the AI model to use
    }
  },

  // 2) THE PHONE LINES — how you talk to it (Step 4)
  "channels": {
    "telegram": {
      "enabled": true,
      "botToken": "123456:ABC-...",               // from BotFather
      "accounts": {
        "+15551234567": { "role": "admin", "name": "Alex" }  // who is allowed to use it
      }
    },
    "discord": {
      "enabled": true,
      "token": "MTIz...",                         // from Discord Developer Portal
      "applicationId": "123456789...",
      "guilds": { "YOUR_SERVER_ID": { "channels": { "CHANNEL_ID": {} } } }
    }
  },

  // 3) MANNERS — when it should reply
  "messages": {
    "directMessage": { "autoApprove": true },     // always reply in private chats
    "groupChat": { "requireMention": true }       // in groups, only when @mentioned
  }
}

• One account, many models — no need to sign up with each AI company separately.
• Private by design — Venice says it doesn't store or log your prompts on its private models.
• Switch any time — change one line in openclaw.json to use a different model.
• Pay as you go — buy credits (100 credits = $1) and only pay for what you use.

1. Create your Venice account through this link (email or Google) so the $10 Pro bonus is added to your account.
2. Open venice.ai/settings/api.
3. Click Create API Key and give it a name, e.g. "OpenClaw server".
4. Copy the key right away — it starts with vnc- and is shown only once.
5. Save it in a password manager.
6. Add some credits — $5 is plenty to start with DeepSeek Flash.

Way 1 — the setup wizard (easiest):

docker exec -it openclaw openclaw onboard
# In the wizard:
# 1. Choose "Venice AI" from the list of providers
# 2. Paste your key (vnc-...)
# 3. Pick the default model (start with deepseek-v4-flash)
# 4. Confirm

Way 2 — edit the settings file by hand:

nano /root/.openclaw/openclaw.json
# Find "providers" and set:
"providers": {
  "venice": {
    "apiKey": "vnc-YOUR-KEY-HERE",
    "defaultModel": "venice/deepseek-v4-flash"
  }
}
# Save: Ctrl+O, then Enter · Exit: Ctrl+X
docker restart openclaw

1. In Telegram, search for @BotFather (the official bot with a blue check) and open it.
2. Send /newbot.
3. Choose a display name, e.g. Alex Assistant.
4. Choose a username ending in "bot", e.g. alex_assistant_bot.
5. BotFather replies with your bot token → copy it.
6. Optional, for group chats: send /setprivacy, choose your bot, then Disable.

"telegram": {
  "enabled": true,
  "botToken": "PASTE-THE-BOTFATHER-TOKEN",
  "accounts": {
    "+15551234567": { "role": "admin", "name": "Alex" }
  }
}
// "accounts" is your guest list: only these phone numbers can use the bot.
// Use your own number with country code. Then: docker restart openclaw

1. Go to discord.com/developers/applications and click New Application (name it "Alex Assistant").
2. In the left menu, open Bot → click Reset Token → copy the token.
3. On the same page, under Privileged Gateway Intents, switch on Message Content Intent and Server Members Intent.
4. Open OAuth2 → URL Generator. Tick bot, then the permissions Send Messages, Read Message History, Embed Links.
5. Copy the generated link, open it in your browser and invite the bot to your Discord server.

"discord": {
  "enabled": true,
  "token": "PASTE-THE-DEVELOPER-PORTAL-TOKEN",
  "applicationId": "YOUR-APPLICATION-ID",
  "guilds": {
    "YOUR_SERVER_ID": {
      "channels": { "GENERAL_CHANNEL_ID": {}, "REPORTS_CHANNEL_ID": {} }
    }
  }
}
// The bot only answers in the channels listed here.
// To copy IDs: Discord Settings → Advanced → turn on Developer Mode,
// then right-click a server or channel → Copy ID. Then: docker restart openclaw

• Telegram: open your bot → tap Start → write "Hi! Introduce yourself."
• Discord: go to one of your listed channels → write "Hi! Introduce yourself."

The reply should arrive within a few seconds.