Completion is not capability.
We built the first platform that treats professional judgment as a measurable skill.
Every compliance training program in the world measures one thing: whether you showed up. Not whether you understood. Not whether you can apply the knowledge under pressure. Not whether — in the moment that matters — your judgment holds. GRID42 was built to measure what actually counts.
"They completed the training."
A completion certificate tells you someone watched a video or passed a multiple-choice quiz. It tells you nothing about whether they can act correctly under real conditions. Regulators know this. Assessors know this. You know this.
"The audit will find out."
The gap between what your team thinks it knows and what it can actually do is called the perception gap. It averages 20–40 points. The auditor's job is to find it. Our job is to find it first.
"AI is accelerating the gap."
As AI reshapes compliance work, the skills that matter most are judgment skills — the ones machines can't perform and that traditional training has never measured. The window to get ahead of this is now.
Judgment is not a soft skill. It is a measurable primitive.
A judgment skill is the ability to recognize a situation, apply the right framework, weigh competing obligations, and act — under conditions of ambiguity, time pressure, and consequence. It can be defined. It can be mapped to a work role. It can be measured with precision. And it can be trained, level by level, until it is demonstrated.
Context judgment
Recognizing which rules and frameworks apply to this situation, in this environment, for this role.
Consequence judgment
Understanding what happens — to the organization, to individuals, to the broader system — when the decision goes wrong.
Synthesis judgment
Integrating requirements across multiple frameworks (CMMC + ISACA + ISO) into a coherent posture.
Orchestration
Coordinating people, processes, and systems toward a compliant outcome under time and resource constraints.
Accountability
Accepting, documenting, and demonstrating responsibility for decisions made.
A Bayesian model that learns what you know.
Bayesian Knowledge Tracing (BKT) updates its estimate of your competency with every response. Not a score. A probability. It knows when you guessed right and when you actually know. It maps which judgment dimensions are strong and which have gaps — producing a precise picture of readiness at the work-role level, not an aggregate score that hides what matters.
role → Work Role 612 · Systems Security Analyst
domain → CMMC L2 · Awareness & Training
p_mastery_context → 0.42 ↑ HAS_GAP
p_mastery_consequence → 0.71 ✓ HAS_MASTERED
p_mastery_synthesis → 0.58 ↑ HAS_GAP
credential_issued → null · gaps detected
routing → pursuit.cubelet.ai · AT.context.L2
Not 'compliance training.' Judgment pathways for specific roles.
Work Role 612 — and every other work role in the NICE Cybersecurity Workforce Framework — has a defined set of tasks, knowledge, and skill requirements. GRID42 maps judgment skill primitives to work roles, generating a precise training pathway for each person based on who they are and what they are responsible for. Not generic awareness training. A judgment pathway built for this person, in this role, at this organization.
The compliance landscape now includes AI agents. Their credentials should too.
For the first time, the same credential infrastructure that certifies a human professional can authorize an AI agent — scoped to specific tasks, bounded by specific domains, verifiable by automated systems. When a regulator, a board, or an orchestration layer asks for evidence of authorized judgment: it exists. Cryptographically signed. Permanently attached to the work performed. Not a certificate. A record.
- Scenario records
- Domain scores
- Human reviewer notes
- JWT-signed
- Wallet-ready
- Task scope
- Domain bounds
- Adversarial test results
- Authorization expiry
- Issuer DID
What is professional judgment in compliance?
What is Bayesian Knowledge Tracing (BKT)?
What is Work Role 612 in the NICE framework?
Can AI agents hold verifiable credentials?
How long does a Scaffold diagnostic take?
See it in action.
The Scaffold diagnostic is running now at app.cubelet.ai. In 20 minutes, you will know exactly where your judgment gaps are — by dimension, by domain, and by work role.