Isolated container
Each agent runs in its own isolated container, with no Docker socket and no access to the host.
Choose a product
Web HostingWebsites and WordPress on your own domainCloudDeploy apps, APIs and databases from Git or DockerAI BuilderCreate and change sites and apps with AIWordPress AI BuilderChange your WordPress site by chattingAdd to any project
DomainsEmailDatabasesManaged WordPressAgent runtimesMigrationsDevelopers
Everything AIDevelopersMCP for AI assistantsDocumentationSecurityTemplatesHostingsurge
PricingSign inAI agents · Coming soon
Private agents. Always on, never a server to run.
When agents open, you will create a Hermes Agent or an OpenClaw agent, connect your own model key and reach it on Telegram or Discord. It will run isolated and always on, with a workspace that is backed up every night.
Coming soon: private agents are not open yet, so none can be created today. This page shows what they will do. The Free plan does not include agents.
What every agent comes with
One managed agent with the features below. You manage it from the dashboard, not from a server console.
Persistent workspace
Its memory, sessions and files survive restarts. The workspace is backed up every night.
Private secrets
Your model key and bot tokens are stored encrypted and never shown again.
CPU and RAM limits
Each agent runs under fixed CPU and memory limits. The dashboard shows them before you create it.
Nightly backups
The workspace is included in the nightly backup.
Telegram and Discord
Talk to your agent on Telegram or Discord. It only answers the user ids you allow. WhatsApp comes later.
Always on
Agents run around the clock and never go to sleep when idle.
Bring your own key
Connect OpenRouter, OpenAI, Anthropic or any OpenAI-compatible endpoint with your own API key.
Two agents, each pinned to a reviewed release
Hermes Agent and OpenClaw are the two agents planned for when agents open. Each will run one reviewed upstream release; moving to a newer one is a deliberate change.
Hermes Agent
Open-source agent from Nous Research (MIT license).
- Version v0.21.6
- Keeps its memory, sessions and skills in its workspace
- Reachable on Telegram and Discord
- Runs under CPU and memory limits
OpenClaw
Open-source assistant gateway (MIT license) that connects an AI agent to your chat channels.
- Version 2026.10.1
- Connects the agent to Telegram and Discord
- Keeps its state in its workspace
- Runs under CPU and memory limits
Coming soon: Neither agent can be created yet. WhatsApp is listed but comes later, because it pairs by scanning a QR code.
From choice to running agent
Five steps in the dashboard once agents open, none of them a server console.
- Choose an agentPick Hermes Agent or OpenClaw and give it a name.
- Connect a modelUse your own provider key, or connect one later. The key is stored encrypted and never shown again.
- Add a channelConnect Telegram or Discord with a bot token and the user ids that may talk to the agent.
- Let it runThe agent is set up and runs around the clock. The dashboard shows its status and health.
- Manage itStart, stop or restart it, read its logs and change its settings. Deleting it keeps one final backup for 30 days.
Isolated by design
Every agent runs in its own container with fixed limits. These boundaries belong to the platform, not to settings you have to get right.
Never given to an agent
An agent gets only its own container and its own workspace.
- The host Docker socket
- Access to the host
- More CPU than its limit
- More memory than its limit
- Another team's workspace
What an agent gets
- RuntimeIts own isolated container
- WorkspacePersistent, backed up every night
- LimitsCPU and memory limits
- SecretsEncrypted, never shown again
- ChannelsTelegram and Discord, allowlist only
- UptimeAlways on
- ModelYour own key
Not a VPS: an agent is a managed container, not a virtual machine you administer. You use it from the dashboard and never administer a host.
Planned prices
Indicative add-on prices for when agents open. Nothing can be bought, and nothing is charged now.
| Plan | Billing commitment | Total billed upfront | Monthly equivalent | Renewal benchmark |
|---|---|---|---|---|
| 24-month plan | 24 months, paid upfront | $186.96 once, for 24 months | $7.79 / month equivalent | $15.59 / month benchmark |
| 12-month plan | 12 months, paid upfront | $124.68 once, for 12 months | $10.39 / month equivalent | $15.59 / month benchmark |
| Monthly | Rolling monthly, no upfront period | $14.29 each month | $14.29 / month | $15.59 / month benchmark |
Prices in US dollars. Nothing is charged now.
Read this before comparing prices: long-term plans are billed upfront for the whole period, so the monthly equivalent is not a monthly charge. Renewal is benchmarked at $15.59 per month. Resource limits are caps, not dedicated CPU or RAM.
Questions worth asking
Straight answers about what a managed agent is, and what it is not.
Do I administer a server?
No. You create and manage the agent in the dashboard: start, stop, restart, settings and logs.
Can I use my own AI key?
Yes. Connect your own key from OpenRouter, OpenAI or Anthropic, or point the agent at any OpenAI-compatible endpoint.
Which channels can an agent use?
Telegram and Discord once agents open, each restricted to the numeric user ids you allow. WhatsApp comes later: it needs pairing with a QR code.
Is this a VPS?
No. An agent is a managed container, not a virtual machine you administer.
What happens when I delete an agent?
The agent and its workspace are deleted. One final backup is kept for 30 days.
Your agent. Isolated and always on.
Coming soon: agents are not open yet.