Nodeau
Sign in Get Nodeau

Nodeau for business

A private AI fleet, on the GPUs your team already owns.

Workstations under desks, a server in the lab, the rack you bought last year. Nodeau turns them into one fleet your team can run models on, with guardrails you set, usage you can see and a record of every change.

Talk to us See the plans

Run it

Every machine, every GPU, every model, in one place.

Each machine reports what it really has and checks its own copy of every model by hash. Nodeau places each workload on a GPU that can hold it, and tells you why it chose that one.

One view of the fleet

Machines, cards and workloads in your account at app.nodeau.ai, from any browser.

Operate it from anywhere

Start and stop models, set scheduling policy, drain a machine, read logs and recover a workload from a machine that's gone.

Maintenance on your terms

Drain a machine before you work on it. New work goes elsewhere and everything already running keeps serving.

Health you can act on

nodeau health shows processor, memory, storage, network and GPU for every machine, a day of history, and alerts. The readings stay on your hardware.

Guardrails

Decide what your fleet may run.

Set your organisation's own limits inside what your plan allows. Set them on a machine with nodeau governance, or from your account in a browser.

  • Quotas. How many workloads, GPUs and batch workers may run at once
  • Model policy. Which models are allowed, by the name they run under
  • GPU pools and fleet groups. Which cards and which machines a workload may use
  • Clear answers. Anything a limit holds back says which limit it was, so people know who to ask
  • Calm changes. Lowering a limit leaves running work alone and shapes what starts next
bash
nodeau governance \
  --max-workloads 6 \
  --max-gpus 4 \
  --allow-model qwen3.5-9b-q4km \
  --allow-model qwen3-embedding-0.6b-q8_0

nodeau governance --json

Visibility

Know what your hardware did, and who changed what.

Usage in accelerator-hours. Which workloads held which GPUs, and for how long, added up across every machine you own. The figure comes from the scheduler's own record of what was reserved. Enter what an accelerator-hour is worth to you and Nodeau does the multiplication.

A record of changes. Your organisation can see what changed, who changed it and what happened, including requests that were turned down. It's kept for a year, and it records actions, never prompts or answers.

an upgrade plan, in short
  target    the release you asked for
  commit    and the exact build behind it

     garage    would move, worker first
     lab       already current
     studio    would move last: the control plane

  Workloads on each machine, and which
  would restart along the way.

Releases

Keep the fleet current, calmly.

  • Know what every machine runs. nodeau update compares each machine with the published release, and a release channel only moves forward
  • Upgrade planning. See which machines would change, in what order and with which workloads, before anything moves
  • Machine by machine. Upgrade each machine with the installer, at the pace you choose

Maintenance windows and a release channel for the whole fleet, set from your account, and rolling the fleet through an upgrade one machine at a time are in progress.

Security

Your data stays on your hardware.

Prompts, completions, embeddings, images and batch records live on the machines that computed them. A connected fleet reports which machines exist, what's plugged into them and what's running, so you can operate it from a browser.

  • Your machines initiate every connection to Nodeau Cloud, keeping the network boundary simple
  • Models keep serving and scheduling keeps working whether or not the cloud is reachable
  • Plans are signed entitlements checked on each machine, and can be carried to a machine with no internet
  • Each machine's agent can write only its own state and reads no credentials
  • Built for machines your team controls, where everyone with access to a machine is trusted

How Business works

Tailored to your fleet.

Every organisation's hardware is different, so Business plans are built around yours: how many machines, which cards, how you deploy and what support you need. Tell us about your setup and we'll put the right plan together with you.

Talk to us

Business brings

  • As many machines and GPUs as you run
  • Everything in Home Pro: batch inference, several GPUs per machine and remote operation
  • Business support, and a direct line to the people building Nodeau
  • A hand planning your rollout

On the way for Business

  • Several people in one organisation, each with a role
  • Single sign-on with your identity provider
  • Keys for scripts that can only ever do less than their owner

The roadmap shows where each one stands.

Building a private AI fleet?

Tell us what you have and what you'd like to run. We'll work with you on the setup, the policies and the rollout.

Talk to us See everything Nodeau does