OpenClaw is a personal AI agent that can actually do things: check your inbox, manage tasks, run commands, and talk to you over WhatsApp or Telegram. Hostinger offers it two ways — as a fully managed plan where you never touch a server, or as a one-click template on a VPS you control. This guide gets you from zero to a working agent with either path, then walks through connecting a chat channel and running your first task.
New to OpenClaw entirely? Start with our overview: Hostinger OpenClaw: What It Is and What It Costs.
What you need before you start
- A Hostinger account with either the AI Automation plan or a VPS — order through our link and the plan provisions automatically.
- A model provider API key (Anthropic, OpenAI, or similar). Create the key in the provider’s dashboard now so you are not hunting for it mid-setup.
- A Telegram account if you want chat control (optional for the first run, recommended right after).
- 15–30 minutes of uninterrupted time. Rushing a setup involving API keys is how secrets end up in the wrong place.
Before you start: pick your path
- Path A — Managed AI Automation plan: Hostinger runs OpenClaw for you. No SSH, no updates, no server config. Best if you want an agent, not a sysadmin hobby.
- Path B — VPS one-click template: you deploy a VPS with OpenClaw pre-installed. More control (custom models, integrations, multiple agents), but you own the maintenance.
Both run the same OpenClaw software. The managed plan trades control for convenience; the VPS trades convenience for control. Price-wise they are in the same neighborhood for one agent.
Path A: Managed setup (about 15 minutes)
- Order the AI Automation plan. The plan covers OpenClaw plus Hostinger’s other automation apps (Hermes Agent, n8n, Paperclip) — you can switch between them. Complete checkout and wait for the provisioning email.
- Open your dashboard. From hPanel, find the AI Automation section and open the OpenClaw control panel. This is where you manage the agent: start/stop, view logs, and reach the web UI.
- Complete the onboarding wizard. You will set an admin password for the web UI and choose your AI model provider. OpenClaw needs a model API key (from a provider like Anthropic, OpenAI, or similar) — this is billed separately by the model provider, not by Hostinger. Paste the key where prompted.
- Open the web UI and say hello. Log in with the password you set and send a first message like “introduce yourself and list what you can do.” If it answers, the core setup works.
Path B: VPS one-click template (about 30 minutes)
- Deploy the VPS. Order a Hostinger VPS and choose the OpenClaw template from the OS/application list during setup. Wait for the deployment to finish (usually a few minutes).
- Note the credentials. The deployment summary or your welcome email contains the server IP and root password. Save these somewhere safe — a password manager, not a sticky note.
- Connect once via SSH (Terminal on Mac/Linux, PuTTY or Windows Terminal on Windows) to verify the service is running and to set a strong admin password for the web UI. The template’s documentation page lists the exact commands.
- Open the web UI at your server’s IP/domain on the documented port, log in, and add your model API key as in Path A, step 3.
- Point a domain at it (optional but recommended). Add an A record for something like agent.yourdomain.com to the VPS IP, then put a reverse proxy with HTTPS in front of the web UI. Never expose an agent dashboard over plain HTTP.
Connect a chat channel
An agent you can only reach through a browser tab gets forgotten. OpenClaw’s killer feature is living inside the messaging apps you already use:
- Pick one channel to start. Telegram is the easiest: create a bot with BotFather, get the token, and paste it into OpenClaw’s channel configuration. WhatsApp and Discord integrations exist too but involve more steps.
- Restrict who can talk to it. Configure the allowed user/chat IDs so only you (and people you choose) can message your agent. An agent that obeys strangers is a liability.
- Test the round trip. Message the bot “what time is it?” — a reply proves the channel, the agent, and the model are all wired together.
Your first real task: the morning briefing
Ask the agent: “Every weekday at 7am, send me a briefing with my unread email count, today’s calendar events, and the top 3 tech headlines.” This exercises scheduling, email access, and web search in one task — and it is genuinely useful from day one. Grant each permission it asks for deliberately; if a request surprises you, say no and ask what it needs it for.
Safety basics (do not skip)
- Approve sensitive actions. Keep human approval ON for anything irreversible: sending emails, deleting files, making purchases, publishing content.
- Secrets stay in the vault. API keys and passwords go in OpenClaw’s credential store, never pasted into chat prompts that might get logged.
- Start with read-only. For the first week, let the agent read and draft but require your OK before it acts. Loosen up once you trust its judgment.
- Keep it updated. On the managed plan this happens for you; on the VPS, update the template regularly — agent software moves fast and security fixes matter.
Frequently asked questions
Do I need the VPS or is the managed plan enough?
For one personal agent doing everyday tasks, the managed plan is enough and far simpler. Choose the VPS if you want multiple agents, custom model routing, unusual integrations, or you simply like owning the stack.
Does OpenClaw include the AI model, or do I pay extra?
Extra. Hostinger provides the hosting and the OpenClaw software; the AI model itself needs an API key from a model provider, billed by that provider based on usage. Budget a few dollars a month for light personal use.
Can OpenClaw message me on WhatsApp?
Yes, WhatsApp integration is supported, though Telegram is simpler to set up first. Either way, lock the bot to your own user/chat ID before connecting anything sensitive.
What happens if I break the configuration?
On the managed plan, you can reset/redeploy from the control panel. On the VPS, the nuclear option is redeploying the template fresh — which is why keeping your API keys and channel tokens saved outside the server matters.