Governance · Flagship case study

GuardrailAI

Controls the model cannot negotiate away.

A compliance-oriented reference framework that places deterministic checks, execution bounds, circuit breakers, and a hash-chained audit ledger around probabilistic agents.

Inspect repository

Problem

Prompt-only safety instructions are not hard controls. High-consequence workflows need explicit ceilings, refusal paths, bounded execution, and records that can be examined after the model responds.

Architectural decision

Deterministic pre-flight and post-flight checks sit outside the model. Circuit breakers select safe fallbacks, while each transition contributes to a tamper-evident ledger.

Execution topology

01
Request
02
Pre-flight controls
03
Bounded agent work
04
Post-flight controls
05
Audit certificate

Engineered invariants

Deterministic rules run before model execution.
Agent work has explicit mathematical bounds.
Post-flight thresholds can refuse unsafe output.
Circuit breakers degrade to defined fallbacks.
Audit entries are linked by cryptographic hashes.

Verification paths

Guardrail sourceDeterministic policy implementations and enforcement boundaries.Source repositoryFramework, tests, deployment assets, and documentation.

Boundaries stated plainly

— A technical reference is not a certification or legal compliance determination.
— Rules must be reviewed and maintained by qualified domain owners.
— Hash chaining supports tamper evidence; it does not replace access control or retention policy.

Enterprise adoption path

01Validate rules with domain counsel.
02Integrate authoritative policy sources.
03Add access control, retention, monitoring, and incident procedures.