Get early access
← All products
AI HARNESS / PRIVATE PREVIEW

Your AI harness. Deployed and in view.

Deploy your organisation’s AI harness and follow its state, health, configuration, and runs from NuPaas.

Discuss your project
HOW AI HARNESS HANDLES AGENTS

Give your AI agents
a place to run.

An agent needs more than a model call. AI Harness coordinates the execution loop. NuPaas brings deployment, model access, and operational visibility around it.

THE STARTING POINTYour repositoryTarget repo + ref
AI HARNESS / AGENT LOOP
MakerWork on the task
CheckerCheck the work
Model requests via NuPaas inference
IN VIEWRun progressTasks · Usage · Outcomes

Conceptual workflow · Available workflows and model integrations are confirmed during private-preview onboarding.

01 Agent orchestration

Run the loop, not just the prompt.

Start a repository-based agent run. The harness coordinates maker and checker work; NuPaas gives you a place to deploy it and follow its operation.

02 Model routing

Connect agents to inference.

Maker and checker requests go through the NuPaas inference gateway, using your organisation’s credentials and configured model access.

03 Token budgets & run control

Keep a handle on execution.

Configure a token budget, inspect reported usage, and cancel a run. Organisation inference quotas are checked before a run starts.

04 Agent observability

See progress and outcomes.

Follow run state, iterations, and task progress. Inspect reported results and halt reasons alongside the harness’s health and configuration.

Harness ≠ model gateway. The harness runs the agent workflow. The inference gateway routes its model requests. Together, they connect execution with model access.

Explore AI inference ↗
HOW IT WORKS

From repository to agent run.

  1. 01

    Set up your harness

    Work with us to configure your organisation’s harness during preview onboarding.

  2. 02

    Follow deployment

    See provisioning state and the health reported by the platform.

  3. 03

    Inspect its operation

    Review configuration versions and the run information available to NuPaas.

Illustrative workflow with example values · Setup confirmed during private-preview onboarding.

WHAT YOU CAN MANAGE

Deploy your harness. Follow its operation.

  • Deployment state and reported health
  • Applied and desired configuration versions
  • Harness runs and operational context
How is AI Harness different from AI inference?

AI Harness is the workload you deploy and operate. AI inference is a separate gateway for model requests, provider keys, and usage. We’ll confirm the available harness workflows and integrations during preview onboarding.

How do I get access?

NuPaas is in private preview. Apply as a founding partner and we’ll discuss your workload, confirm availability, and help with onboarding.

What does it cost?

We’ll discuss preview access and founding-partner terms directly. Public pricing will follow as the launch offer takes shape.

FOUNDING PARTNER PROGRAM

Build something real.
Help shape what’s next.

We’re opening the conversation with our first 10 founding partners. We’re looking for teams with a real workload, a clear use case, and time to share feedback. Tell us a little about yours.

  • Early access with onboarding support
  • A direct conversation with the builders
  • Your use case helps shape the platform

We review each application and follow up to discuss fit.