cloudrun

What It Does

cloudrun is a managed hosting platform designed specifically for deploying and operating AI agents in production. It serves as a control plane that provisions isolated infrastructure for each agent, handles configuration, and provides a unified interface for managing entire agent fleets. The platform currently supports Hermes agents with additional runtimes planned.

Who It's For

The platform targets three primary audiences:

• Developers who have built AI agents and need reliable cloud hosting with tool connections and model provider integrations
• Automation engineers looking to schedule recurring agent tasks and connect business applications
• Agencies delivering agent solutions to clients who need multi-tenant project management and clean handoff workflows

Key Features & Capabilities

• Isolated Infrastructure: Every project gets its own dedicated machine with separate blast radius
• Pre-Baked Provisioning: Agents deploy from golden snapshots with runtime and browser already configured, not built at provision time
• Tool & Channel Connections: Connect services once at the agent level — credentials are held securely and never pasted into prompts
• Multi-Channel Support: Agents work where you do — Telegram, Slack, email, and more
• Scheduled Automations: Run prompts on cron timers with automatic reporting
• Unified Canvas: Manage organization, projects, engines, and agents from one interface instead of per-agent dashboards
• Flexible Model Providers: Start with free tier models, connect your own API keys when you need more power
• Safe Billing: Payment is charged before provisioning so declined cards never leave running infrastructure behind

Typical Workflows

Developers configure their Hermes agent through the dashboard, connect necessary tools and model providers, then test workflows before going live. Automation builders define tasks, connect supported apps, and schedule recurring operations. Agencies set up isolated client projects, validate functionality, and prepare handoff documentation.

What Makes It Different

Unlike traditional agent runtimes, cloudrun is a control plane that provisions dedicated infrastructure per project rather than running agents in shared environments. The platform uses pre-baked snapshots for instant provisioning and charges only for running deployments (starting at $49/month, dropping to $39 for additional agents). Projects that lapse are paused and retained for 30 days rather than immediately deleted. The architecture is designed to accommodate multiple engine kinds beyond Hermes, with OpenClaw, Codex, Claude Code, OpenCode, Pi, and Cursor already wired for future release.