Create your organization
Continue with Google on the sign-in page and CloudOps creates your organization and its tenant. Connect servers you own, or rent capacity from the marketplace.
AI DevOps on your own hardware
Build your private cloud.
Bring your AI agent.
CloudOps is the control plane for servers you own or rent. Connect Claude or any assistant that speaks MCP, and it can inspect your capacity, propose VMs and volumes, and make changes only within the permissions you choose.
Sign up with Google. A person creates the account and approves agent access before any agent connects.
- Own or rent bare metal
- Permissions you choose
- Every tool call recorded
- Help from a person
How setup works
From hardware to an agent-managed private cloud.
Turn on AI agents
An organization owner allows agents for the tenant under Settings → AI agents on your tenants. In the tenant console, open Developers → AI agents (MCP) and choose Read only or Read and safe changes.
Start with one test resource
Ask your agent to inspect capacity and propose one small VM or volume. Approve the plan, then check the result and the job in the console.
Creating resources is a separate switch: “Also let agents create VMs and volumes, and delete what they created”. VMs and volumes are billable, and the organization’s limit always applies.
Bring your existing assistant
Point your agent
at CloudOps.
Use any assistant that supports remote MCP servers. In Claude.ai or Claude Desktop, open Settings → Connectors → Add custom connector, paste your tenant endpoint, then sign in to CloudOps and approve the access.
Your exact endpoint is shown in the tenant console under Developers → AI agents (MCP). Other MCP clients can authenticate with a personal API token instead.
Tenant MCP endpoint
https://<your-tenant>.cloud-ops.io/api/mcp
The organization endpoint, https://cloud-ops.io/api/mcp, only reads organization details. It cannot change infrastructure.
Start with this prompt
Help me set up a small private cloud with CloudOps (https://cloud-ops.io) on hardware I own or rent. Walk me through creating my account and tenant. Ask me for my tenant MCP endpoint; I will sign in and approve the access. Inspect the tenant's tools, permissions, servers, images, networks and storage. Propose one test VM or block volume. Explain the capacity, the cost and any missing prerequisites, and wait for my approval before any billable change. Use only the tools and permissions that are available. Report the resource and job status, and tell me how to review the activity and remove the test resource.
Paste this into your assistant once it is connected.
Know what can change
Permissions you can inspect.
Can my agent build the whole cloud through MCP?
Not on its own. Agents can create VMs and block volumes. A person creates the account, connects or rents the hardware and sets up the prerequisites. Other infrastructure is managed in the console, the REST API or the Terraform provider. Ask the agent to list its tools before it makes a plan.
Can I start with read-only access?
Yes. Read only lets the agent look at servers, networks, storage, clusters, databases, alerts and jobs without changing anything. A tenant admin can lower the access or turn it off at any time, and it can never exceed what the organization allows.
What can an agent delete?
When creating is turned on, an agent can delete only the VMs and volumes its own connection created. It cannot delete anything a person or another connection made, and it cannot change firewall rules, users or access settings.
Where do I review activity and get help?
Developers → AI agents (MCP) in your tenant lists every tool call. Resources and their jobs are in the console. For setup help, talk to the CloudOps team.
Make your first agent-assisted cloud plan.
Start with one tenant and one test resource, and keep control as you grow.