Build and test software
Give a coding agent a workspace for repositories, dependencies, builds, and tests. Keep the job on its own machine while you use yours.
Nova Agent Computers / Pilot inquiries
Give your agent
a Linux computer.
A place to code, use a browser, and run tools. Bring your agent. Tell us what it needs. We’ll scope a Linux workspace on Hungry Nova Labs hardware, with computer time priced separately from model usage.
We’re taking inquiries for a small, manually provisioned pilot. Configuration, availability, and pricing are confirmed before setup.
Hardware we own in AtlantaBring your own agentDiscuss your workload with us
What it’s for
Your agent has work to do. Give it a place to do it.
Run browser workflows
Scope a Linux browser environment for research, QA, and repetitive web tasks. Tell us whether your workflow needs a visible desktop or a headless browser.
Keep a working environment
Need the same files and installed tools next time? Ask for retained storage. We agree on retention, backups, and export before the pilot starts.
The pilot
Start with one agent and one useful job.
Tell us what needs to run
Share the agent, tools, expected hours, and any browser or storage requirements. We confirm the machine size, rate, and a runtime limit.
Get a workspace set up with us
The pilot starts with manual provisioning and private access. SSH and desktop options depend on the agreed configuration. Bring your own model account and credentials.
Use the computer for the agreed time
Run your workload and review usage with us. Stopping compute and keeping files are separate decisions; both belong in the quote.
A clear bill
Computer time.
Model usage.
Separate costs.
The offering is built around time-based rental. We quote the pilot before setup, including how time is measured and when billing starts and stops.
Computer cost = running time × agreed machine rate
A running machine reserves resources even while its agent is waiting. Saved storage, setup, and any other charges are listed separately in your quote.
- Bring your own model provider; its token charges stay with that provider.
- Agree on a runtime window and spending limit before starting.
- Ask about a retained workspace for recurring jobs.
Self-service checkout, prepaid compute credits, and automatic start/stop controls are planned. They are not available on this page.
Where we’re going / Planned
Find a computer.
Rent it.
Get to work.
We’re building toward rentals that an agent can arrange through an API: check available machines, get a quote, pay from an authorized budget, and receive temporary access.
Discover → Quote → Pay → Run → Stop
Each rental is intended to have a fixed expiration, a spending cap, and limits on repeat rentals. A configurable cooldown gives operators control over how often an agent can start another session.
This is the roadmap, not a live rental API. Today, start a pilot conversation with us. Automated payments and provisioning will follow validation of capacity, billing, and isolation.
Before you start
A few practical details.
Am I renting a computer or buying an AI model?
You’re discussing a Linux computer rental. Your agent runs its tools there and can call the model provider you choose. Model inference and GPU capacity are not included unless explicitly quoted.
Can my agent rent a machine on its own?
Not yet. The pilot is arranged with a person through the contact form. API access, scoped agent credentials, and prepaid spending controls are part of the planned self-service service.
Is this a whole physical server?
The starting offer is a Linux virtual machine on Hungry Nova Labs hardware. CPU, memory, storage, and resource-sharing arrangements are confirmed in the quote. Dedicated hardware is a separate conversation.
How will rental machines be isolated?
The rental service is planned for its own VLAN, separated from management and existing customer services. Individual rentals also require isolation from one another. These controls must be validated before unattended rentals launch.
Will my agent and tools work?
Tell us which agent you use and what it needs to install, access, and run. We check fit before provisioning. Browser control and remote desktop support depend on that setup.
Do files disappear when the computer stops?
That depends on the retention option agreed for your pilot. We specify what is saved, any storage charge, the export process, and when data is removed before you start.
Does everything stay on Hungry Nova Labs hardware?
The computer runs on our hardware. If your agent uses an external model or website, the data it sends goes to that service. We can discuss local model requirements separately.
Let’s scope your first workspace
What would your agent do
with its own computer?
Tell us the agent, the job, and roughly how long it needs to run. We’ll work out the configuration and a pilot quote with you.
Send a Message