Principle

Let the model propose; let deterministic code commit

Use the model for judgment; use the control plane for authority.

When it fits

  • An agent can trigger consequential mutations directly from model output.

When to avoid it

  • Deterministic guards constrain effects but cannot guarantee the model chose the right business objective; consequential intent still needs appropriate review.

Why it matters

Treat model output as a proposed action. Deterministic infrastructure validates schema, permissions, limits, idempotency and current state before committing the mutation. Record the decision and result so failures can be audited, retried or reconciled without asking the model what probably happened.

An example

An agent proposes updating 500 customer records. The control plane validates the allowed fields, batch size and authorization, assigns an operation ID and performs the write.

Check your result

A malformed or unauthorized proposal fails before mutation, and a later operator can reconstruct why an accepted action ran.

Keep this limit in mind

  • Deterministic guards constrain effects but cannot guarantee the model chose the right business objective; consequential intent still needs appropriate review.

Connected ideas

Useful with
Enforce permissions outside the model

Evidence and sources

Supports

A deterministic control plane can treat model outputs as proposals, validate them, commit permitted actions and keep a decision history for recovery and audit.

Deterministic infrastructure cannot make model judgment deterministic; it constrains effects and records decisions rather than eliminating model uncertainty.

Building Deterministic Infrastructure for Non-Deterministic AI Agents · 3:08-6:27, model proposals, control-plane decision history, coordinated state and distributed-systems protections

All sources (1)