Launch OpenClaw.Keep your AI online.

OpenClaw is an open-source AI assistant that connects a model to your tools, files and messaging channels. Its Gateway runs on your computer or server and coordinates the work you assign.

Unlike a chat window on its own, it can use configured tools to work with files, browse websites and run scheduled tasks. You choose its model, connections and permissions.

Depending on the tools you connect, OpenClaw can:

• Coordinate coding work and prepare changes for review.
• Research topics and draft documents.
• Process email and create tasks in connected systems.
• Carry out browser workflows and scheduled checks.
• Return results through supported messaging channels.

These workflows need configuration. See the community examples and tasks you can try.

The OpenClaw software is free and open source under the MIT licence. You can install it without buying a software licence.

Running it can still involve hosting, electricity, AI model usage and paid integrations. Truehost’s plan price pays for hosting; it is not an unlimited AI-credit allowance. See the OpenClaw licence.

Choose a local computer or hosted VPS, install OpenClaw, then complete its setup:

1. Connect a supported AI model provider or model endpoint.
2. Open the dashboard or connect a messaging channel.
3. Add the tools and accounts needed for one task.
4. Test the task and review its output before adding automation.

Truehost’s OpenClaw plans include a prepared installation. You configure your model and integrations. Choose a hosting plan.

No GPU is needed for the Gateway when your AI model runs through a cloud provider. The provider handles model inference; your computer or VPS runs OpenClaw and its tools.

If you also want to run a model locally, check that model’s RAM, GPU or unified-memory requirements. A GPU may improve inference speed, but there is no universal GPU specification for every model. These hosting plans are not advertised as GPU servers.

The official hosting guidance lists 1 GB RAM as an absolute minimum and 2 GB or more as recommended headroom for a basic Gateway.

Allow more for browser automation, media tools and concurrent tasks. These figures do not include the memory needed to host a language model. Monitor actual usage when choosing between the plans.

If you mean disk space, there is no single fixed size for a complete setup. The official basic-host guidance lists roughly 500 MB of disk space, but that is not a total storage budget.

Allow additional space for the operating system, dependencies, browser downloads, logs and workspace files. Local model downloads add their own storage requirements. Package size also changes between releases.

Yes. OpenClaw supports local setups on macOS, Linux and Windows. Your computer must remain awake and connected for ongoing tasks.

A VPS keeps the Gateway independent of your laptop. You can also pair a supported local device with a remote Gateway for tasks that need that device. Compare the local, hosted and hybrid options.

Start with a compatible computer you already own if you will use a cloud AI model. A recent laptop, desktop or low-power mini PC can be suitable; you do not need to buy a particular brand just to run the Gateway.

For a dedicated setup, prioritise reliable power, SSD storage, cooling and enough spare memory for your tools. Our practical suggestion is 8 GB RAM for light desktop use, or 16 GB if you also run browsers and development tools; these are planning recommendations, not official minimums.

If you want local model inference, select the model first and size the hardware around it. There is no single best computer for every model and workload.

For a dedicated server, Ubuntu LTS is a straightforward starting choice. OpenClaw’s hosting guidance identifies modern Debian/Ubuntu as its best-tested Linux installation path.

For local use, macOS, Linux and Windows are supported. Windows offers native setup options and WSL2 for a Linux-style environment. Choose the platform that supports the tools you need and that you can maintain comfortably. See the current installation options.

A VPS can be a suitable environment, but hosting alone does not make an agent safe. OpenClaw can act through the credentials and tools you give it.

• Keep administrative access private and authenticated.
• Limit tool permissions and use isolated execution where appropriate.
• Review third-party skills before installing them.
• Keep software updated and maintain recoverable backups.
• Require review for consequential actions, such as publishing or sending payments.

Follow the official security guidance. Separate clients or users who should not share access.

For occasional use, an existing computer usually avoids an additional hosting bill. Total cost still includes electricity, model usage and any paid tools.

For ongoing automation, compare a VPS subscription with keeping a computer powered on. Choose a cost-effective model suited to the task, limit unnecessary scheduled runs and track provider usage. Compare Truehost’s current plans for a prepared hosted setup.

Use it to help deliver a service customers already need. Possible offers include setting up business workflows, maintaining client automations, supporting website development, or producing research and reporting deliverables.

Start with a specific outcome, such as turning incoming enquiries into organised follow-up tasks. Agree on scope, test the workflow and charge for implementation or ongoing maintenance.

Include hosting, model usage, support time and quality checks in your costs. OpenClaw itself does not generate guaranteed income; your expertise, customer demand and delivery determine whether the service is profitable.

Both offer tool-using agents, messaging access and scheduled work. OpenClaw centres its experience on a persistent Gateway connecting channels, workspaces and devices.

Nous Research positions Hermes Agent around a built-in learning loop: creating and refining reusable skills, persistent memory and recalling prior sessions. It also supports CLI and messaging workflows.

Compare the integrations, permissions, deployment options and model support you actually need. Test the same representative task in each before choosing; neither is universally better. See OpenClaw’s documentation and the official Hermes Agent project. This page’s plans are for OpenClaw.

Yes, the listed hosting plans include pre-configured OpenClaw and automated installation. You still connect your model credentials, channels and tools.

Choose a plan based on the resources your workflows need, then complete your workspace setup.

The Gateway and server-based tasks can keep running on your VPS. They still depend on the server and connected services being available.

If a task uses a paired local device, that device must be online. Hosting does not automatically give the server access to your laptop’s files or applications.

A compatible setup can be migrated. Back up the workspace, configuration and state, then restore and test them on the new server.

Check file paths, credentials, channels and scheduled jobs before retiring the old installation. Follow the migration documentation.

The plans list one-click updates. Keep a recoverable copy of your workspace and settings before making changes.

Confirm the ongoing operating-system, application and backup responsibilities included with your service. Test restores and check connected tools after updates.