AI governance you can prove, not just promise.
Guardrail is an AI governance consulting practice for organizations deploying AI under real scrutiny — from regulators, auditors, boards, and their own customers. We build the controls, oversight, and evidence that let you show your AI does what you say it does.
For the people who have to prove it. Guardrail is still forming and is not yet accepting client engagements.
What Guardrail does
Most AI failures aren’t model failures — they’re governance failures: no inventory, no owner, no evidence. Guardrail works on that layer.
AI risk & controls assessments
An inventory of the AI systems you actually run — including the shadow AI nobody approved — mapped against the risks each one carries, with the controls that should be watching them.
Governance frameworks & policy
Decision rights, approval paths, and policies that survive contact with real teams: who can deploy what, who reviews it, and who can stop it.
Evidence & audit readiness
Documentation, testing records, and oversight logs built to hold up in an audit — so “trust us” becomes “here is the evidence.”
AI literacy for teams
Training built on the Five-Pillar AI Literacy Standard, so the people using AI every day know the system, its failure modes, and where its boundaries are.
The Five-Pillar AI Literacy Standard
Guardrail’s work is organized around a simple standard: five things every team that touches AI should be able to do, resting on one foundation.
Know Your System
What the system is, what it was trained and built to do, and where it sits in your decisions.
Know the Failure Modes
How it breaks, hallucinates, drifts, and gets misused — before your customers find out.
Exercise Human Oversight
A named human who reviews, can override, and actually does — not a rubber stamp.
Respect the Data
What goes in, where it goes, who can see it, and how long it stays there.
Know the Boundaries
The uses the system was never meant for — and the discipline to refuse them.
The standard is published in interactive form alongside AI, Audited, the practice’s newsletter on AI governance and accountability.
About the practice
Guardrail LLC was founded by Glorin Sebastian, an IT risk and controls lead and doctoral researcher whose work sits where AI systems meet financial reporting, audit, and accountability.
His day-to-day discipline is the unglamorous core of governance: completeness and accuracy of the reports systems produce, tie-outs that reconcile, and evidence that an independent reviewer can re-perform. Guardrail brings that same standard to AI.
He also publishes AI, Audited, a newsletter on AI governance and accountability — written for the people who have to prove it.