Most teams start agents on a laptop or a shared server. That works until the machine sleeps, the process dies, or one client’s agent can see another’s files. Cloudrun is built for the opposite: one box per project, a baked runtime snapshot, and a fleet view so you can see every agent in one place.

Product docs and the product itself live at usecloudrun.com and docs.usecloudrun.com.

What “isolated cloud machine” means here

Each project gets its own isolated box. The agent’s shell, files, and runtime live on that machine, not on your laptop, and not on a shared host with other projects.

A baked runtime snapshot keeps the environment consistent when the agent starts. You are not reinstalling tools by hand every time the process restarts.

Fleet canvas: one place for many agents

When you run more than one agent (dev experiments, production bots, or client work), you need a single canvas for the fleet. Cloudrun’s fleet canvas is that view: which agents exist, how they are connected, and how they are hosted, without jumping between unrelated dashboards.

Connectors and channels agents actually use

Agents need tools and inboxes, not just a chat UI. Cloudrun supports connectors including Gmail, Notion, Linear, GitHub, Telegram, Google Calendar, Google Sheets, Jira, HubSpot, Discord, Shopify, and Zendesk.

For talking to users and schedules, multi-channel options include Telegram, Slack, and email, plus cron for timed runs.

Who this is for

Developers who want an agent with real shell access on a machine that stays up when their laptop does not.
Automation builders who need longer runs, connectors, and cron, not a one-shot function timeout.
Agencies hosting client agents who need separation between clients, not one shared box for everyone.

How this page connects to common questions

These child pages go deeper on the complaints people actually hit:

Next step

Deploy and manage agents on isolated machines at usecloudrun.com.

Deploy an AI agent to an isolated cloud machine