02 / PLATFORM

Loose on reasoning. Rigid on control.

Light is the reusable execution and governance foundation beneath the solution. It governs the actions surrounding a probabilistic model; it does not make the LLM deterministic.

The execution boundary

Keep model choice separate from authority.

Local, enterprise-hosted and frontier models can support agents. Light governs identity, authorization, scope, approval, tool access, state, audit, execution and reconciliation before those agents interact with your systems.

Your environment. Clear boundaries.
Local / private AIEnterprise AIFrontier APIs
Agents · reasoning & preparation
LIGHT · GOVERNED EXECUTION
  • Identity
  • Authorization
  • Scope
  • Approval
  • State
  • Audit
  • Policy
  • Tool control
  • Reconcile
Your systems, data & workflows

Reasoning can be flexible. Actions need permission.

An illustrative architecture, adapted to your infrastructure and operating model.

Control fields + business context

Flexible data. Unambiguous control.

Customer-specific information should not require rebuilding the whole platform. Combine known control fields, flexible business payloads, evidence, reasoning instructions and structured output.

  • Controlled structured fields

    Authorization, routing, enforcement, thresholds and state transitions use explicit structured data. A sentence in a business document does not grant permission.

  • Flexible business payload

    Documents, text and JSON carry domain-specific context. Keep their evidence and reasoning instructions separate from trusted control fields.

  • Validated output

    Check structured output against the workflow contract, then apply the human or policy boundary before returning it to an existing process.

Model placement

Use the model that fits the workload.

Choose according to security, capability, performance and cost. Governance stays consistent across providers; no model is universally superior.

  1. Local / private models

    A placement option for sensitive data, subject to the organization's data controls and operating capacity.

  2. Enterprise-hosted AI

    Controlled serving and access for workloads within your enterprise platform.

  3. Frontier APIs

    An option for difficult reasoning when data policy permits the external service.

  4. Smaller models

    Evaluate focused models for classification, extraction and routine processing.

An optional infrastructure fit

Nutanix Enterprise AI + Light.

Nutanix Enterprise AI can provide the model and inference infrastructure. Light provides governance, workflow and execution integration. This is an architectural integration option, not an official endorsement, Nutanix Ready designation, partner product or joint-support claim.

  1. Nutanix Enterprise AI

    Models / inference infrastructure

  2. Light

    Governance / workflow / execution

  3. Enterprise systems and data

    Your existing environment

Start with the system. Tell us what needs to work.

An infrastructure decision, a migration, or a workflow you want AI to help with. Bring the problem, the data you have and the outcome you need. You do not need to choose a model or framework first.

Discuss your use case