Run AI agents without expensive GPU infrastructure
Keep your AI agents, automations and workflows running 24/7 on cloud infrastructure instead of relying on your laptop or local environment. Run Hermes, n8n, lightweight Ollama workloads, Claude-powered agents, scrapers, and other agent frameworks on your own Linux VPS with full root access.
Agents you can run on Webdock
- n8n
Hermes Agent
OpenClaw- Ollama
Model APIs your agents can call
- OpenAI
- Anthropic
Grok- OpenRouter
Run your AI agents 24/7 on a VPS
Running an agent on your laptop gives it access to your local files, and it stops when the laptop sleeps. A separate Webdock VPS gives it an isolated environment that runs 24/7.
Run 24/7
Your agent keeps working when your laptop is closed, offline or asleep.
Full Linux Environment
Install the frameworks, packages, Docker containers, and dependencies you need with full root access, including Python hosting with pip ready on first boot.
Persistent Infrastructure
Keep your agent, files, configuration, IP address, and workflows available in one persistent environment.
Fast Network Connectivity
1 Gbit/s networking, up to 5 Gbit/s, for API calls, integrations, scraping, and moving data between services.
Laptop vs VPS
On a laptop the tasks drop out as the machine sleeps. On a VPS the same agent keeps running.
Agent on your laptop
- Scheduler tickactive
- Tool callsactive
- Webhook listeneractive
- Memory / stateactive
Agent on a Webdock VPS
- Scheduler tickactive
- Tool callsactive
- Webhook listeneractive
- Memory / stateactive
What Webdock does and what you do
What Webdock does
- Keep your server running 24/7 on dedicated resources
- Give you full control with root access, even from a terminal in your browser
- Give it a fixed IP address, so webhooks and integrations always find it
- Back it up for free, daily and weekly
- Protect it from DDoS attacks
- Help you on live chat, phone and email
What you do
- Install your agent, for example with Docker
- Add your own API keys, for example OpenAI, Anthropic, Grok or OpenRouter
- Keep it updated, on your own schedule
Example agent setups
Morning research reports
Built with
What it does: Wakes up every morning, pulls data from the APIs and sites you choose, summarises it through OpenAI, Anthropic or OpenRouter, and posts the report to your team chat or inbox.
Why a server: It runs at 7:00 whether or not anyone's laptop is open.
Inbox and invoice triage
Built with
What it does: Every 15 minutes it fetches new messages, lets a model classify them, writes the result to your database and notifies the right person.
Why a server: Schedules and webhooks need an address that is always online.
Private document Q&A
Built with
What it does: Answers questions about your internal documents using a CPU-friendly model that runs on the server itself.
Why a server: Your documents and questions never leave your own server, hosted in Denmark.
Webdock MCP
Your agent is running. Now manage its server from your AI.
Webdock MCP lets your AI client look after the VPS your agent lives on. Connect ChatGPT, Claude, Devin or GitHub Copilot, then deploy, resize and snapshot servers in plain language, with permissions you approve. Included with your Webdock account.
AI clients that can manage Webdock with MCP
Webdock MCP
Hand your server work to your AI
Check, resize and snapshot the VPS your agent runs on by simply asking in plain language.
You sign in with OAuth, so no password or API token goes into the chat, and nothing changes without your approval. Webdock MCP is included with your Webdock account at no extra cost.
Step 1
Sign up.
Create your Webdock account.
Sign upStep 2
Add Webdock MCP in your preferred AI.
Add Webdock in Claude or similar, sign in, and choose what it's allowed to do.
See setup guidesStep 3
Start prompting.
Ask in plain language. Your AI selects the right Webdock tools and performs the task. ”Deploy my app on a Webdock server in Denmark.”
See how Webdock MCP works
What runs well on Webdock
Agents using external LLM APIs
Great fitYour agent runs on the VPS and connects to external models through APIs such as OpenAI, Anthropic, Grok or OpenRouter. The heavy AI work happens there.
Small local models through Ollama
Good fitQuantized models in roughly the 7B to 13B range can run on sufficiently sized instances, although performance depends heavily on available CPU, RAM and the model itself.
Large models at high speed
Not yetIf you need to run large 70B+ models locally at high speed, you'll want dedicated GPU infrastructure. That's not us. Yet.
Managing your servers from your AI client
IncludedConnect Claude, ChatGPT, Devin or GitHub Copilot with Webdock MCP to check, resize and snapshot the VPS your agent runs on.
Frequently asked questions
Everything you need to know about our pricing and plans.
No. You install your agent yourself, for example with Docker, and our Linux engineers can help with server questions.
No, not for agents that use external APIs such as OpenAI, Anthropic, Grok or OpenRouter, or for smaller local models through Ollama. Webdock VPS instances are CPU-based.
Anything that runs on Linux. You get full root access, so n8n, Hermes Agent, OpenClaw, Ollama, Docker, Python, Node and your own code all work.
Webdock MCP connects the AI client you already use, such as Claude, ChatGPT, Devin or GitHub Copilot, to your Webdock account. You then manage your servers by typing what you want in plain language, for example "Show me all my servers" or "Take a snapshot before I update."
Your agent runs on the server. Webdock MCP helps you look after that server. Ask your AI to check its status, take a snapshot before an update, or give it more RAM when your agent needs it, without opening the dashboard.
No. If you can write a message in an AI chat, you can use it. You connect once with a secure login, approve what the AI client is allowed to do, and then just ask.
No. Webdock MCP lets your AI client deploy and manage your Webdock servers. AI agent hosting runs your own agents on a Webdock VPS. You can use either one, or both.
No. It's included with your Webdock account. You only pay for the servers you run.