AI guardrails framework / any sector

Set, Prove & Hold

The difference between a rule an AI has been asked to follow and a control designed to resist bypass.

A guardrail is only as strong as the layer it is actually built into.

Set, Prove & Hold (4 + 4 + 4)

Set is sequential before build, Prove supplies four pre-launch tests, and Hold runs four concurrent disciplines after launch.

In order

Set

Steps before you build

Four tests

Prove

Tests before you launch

In parallel

Hold

Disciplines after launch

SetSet action 1 of 4

Name what must never happen

Write a rule someone could try to break this afternoon.

  • Write a "never", not a value: "be helpful and safe" cannot be tested.
  • Write it for this system rather than copying a vendor's generic list.
  • Use the applicable law, the damage this system could cause, and known AI failure modes as inputs.

Where the rule is enforced decides what it can resist

The same customer-data rule can be expressed in four places. The layer changes what reaches the AI and what limitations remain.

One rule, four ways“Never reveal another customer’s data.”

What it is

The AI receives only task-needed data and tools

Customer-data example

Correctly scoped retrieval returns records for the signed-in customer only.

Customer-data flow boundary

Scoping is before the model: only task-needed, authorised records reach the AI.

Customer dataScoped records
scope →
AI modelOnly authorised, task-needed customer records reach the AI.
response →
DeliveryUser or connected system

What reaches AI

Only authorised, task-needed customer records reach the AI.

Limitation

It depends on correct identity, authorization, and scoping design.

Build each rule into the strongest layer that can carry it. Architecture limits exposure; runtime checks add an independent boundary; policy and prompts remain useful but are not equivalent controls.

The layer changes how you build, test, and re-test

Compare four enforcement layers across the Set, Prove, and Hold lifecycle.

LayerSet — build itProve — test itHold — re-test it
1 · PolicyWrite it in a documentReview whether the policy is complete and communicated; it does not itself execute a technical test.Review when the policy or use changes.
2 · PromptPut it in instructionsAttack the instruction and record residual failures.Re-test on model or instruction changes.
3 · RuntimeAdd an external input or output checkAttack it and measure wrong blocks.Re-test when the scanner, model, or supported languages change.
4 · ArchitectureScope data and tools to the taskVerify authorization and scoping are in place.Review when identity, data access, tools, or architecture change.
Context

Layer choice is a maintenance decision as well as a security decision: weaker, change-sensitive controls generally demand more frequent evidence.

Measurement standard

Track the number of active guardrails at each layer, alongside dated test evidence and legitimate-work block rates.

Where to start

Three practical moves, in order.

  • 1
    Inventory the guardrails you think you haveMark the enforcement layer for each and review how many rely only on policy or instructions.
  • 2
    Try to break the three that matter mostUse the deployed system and record the attempts, versions, results, and legitimate-work effects.
  • 3
    Move one guardrail to a stronger layerStart by narrowing the data and tools available to the task.

Turn guardrails into evidence

We can help your team map controls, test the deployed system, and establish a reviewable operating rhythm.

Start a conversation
Related Methodology

Agent Authority Model

Authority decides what a handover may do; guardrails decide how strongly its boundaries are enforced.