TLDR: machine0 is a CLI for agent compute: “machine0 new mybox” gives your agent a full cloud VM, billed by the minute, from $0.013/hr up to 60 vCPU / 240 GB and 8×H100s. Agents drive it themselves via CLI or MCP.

https://www.youtube.com/watch?v=gyllkZ0M04E

The Pain

Agent workloads are moving from ephemeral to always-on. A coding agent working on a hard feature runs 6-8 hours. Auto-research agents run for days. OpenClaw & Hermes run 24/7. As you run more in parallel:

• Resources: a few agents on a big codebase saturate your RAM and CPU. Auto-research needs GPUs you don't have.
• Security: “—yolo” on your personal machine is one prompt injection away from exfiltrated credentials.
• Availability: close your laptop and the agent dies mid-task.
• Isolation: there's no clean line between everything on your laptop and the minimum this agent actually needs.

The Solution

machine0 gives every agent its own computer. It's a CLI that both humans and agents can use without reading docs:

• “machine0 new mybox” creates an SSH-ready VM with a static IP and HTTPS endpoint. Always on (99.99% uptime) until you switch it off.
• Billed by the minute. 1 vCPU / 1 GB at $0.013/hr up to 60 vCPU / 240 GB, plus GPUs from RTX 4000 Ada to 8×H100s.
• Profiles. Bundles of MCP connections, credentials, prompts, and env vars, injected at VM creation. Each agent gets exactly the capabilities (or personality) you choose, and nothing else.
• Agents self-serve. Hand the CLI or MCP server to Claude, Codex, or OpenCode and it manages its own fleet: spin up a box for a build, snapshot it, tear it down.
• Suspend & resume. Pause an idle machine, stop paying for compute, resume exactly where it left off.
• Snapshot & clone. Configure one machine perfectly, then stamp out ten identical agents from it.
• NixOS is first-class. Declarative, reproducible machines.

How people use it

• Compute for your agent: give it the CLI and it spins up its own VMs for builds, tests, and long jobs instead of grinding your laptop.
• Hosting the agent itself: run Claude, Codex or Hermes 24/7. Snapshot once, then clone as many as you want.
• Swarms of agents: an orchestrator agent spawns worker machines, farms out subtasks, and tears them down when the work is done.
• Auto-Research on GPUs: drop an agent on an H100 with a goal and let it run for days.

About me

I'm a solo, second-time YC founder. I was CTO of Upflow (YC W20), where I scaled EPD post series A and personally owned DevOps, infra, and security for years because they were too high-stakes to delegate. Upflow handles sensitive accounting data for companies like Postman, ProductBoard, and Lattice, and is doing many millions in ARR.

Before that I built consumer payments at Hike Messenger (100M+ users) in India. Working with sensitive data and infra for years is what got me deep into the problems machine0 solves, security especially.

Asks 🙏

• Try it: https://machine0.io
• Let's talk about your agent fleets, software factory or auto-research: hello@machine0.io