The 72/10 Problem: Why Industrial AI Stalls Before It Scales

The gap is not technical

Seventy-two percent of manufacturers have deployed AI in some form. Ten percent have scaled it beyond a pilot or single line Automation World. That gap should worry any energy or industrial operator who has budgeted for AI as a productivity lever rather than a demo. The data points to workforce mistrust and poor internal communication as the primary blockers, not model performance Automation World. That finding should reshape how compliance and operations leaders sequence their AI investment.

Systemic AI raises the stakes

The industry’s own framing has moved past isolated pilots toward what Accenture and others call systemic AI, where decision-making systems are directly integrated with physical execution across plants, not bolted onto a single workstation Automation World. That integration is where the risk profile changes. A pilot that misfires costs you a demo. A systemic deployment that misfires touches safety interlocks, output quality, and potentially the grid-facing load management now central to plant operations as data center and industrial power demand reshapes sourcing decisions Cummins via ProcurementMag. Scaling without governance infrastructure is not a productivity risk anymore. It is an operational one.

Knowledge capture is the missing layer

System integrators on the ground are naming the real constraint plainly. Alden Olyniec of Atlas Prediction Control, speaking through the Control System Integrators Association, notes that AI’s value in plants depends on capturing institutional knowledge and keeping that data secure, not on the sophistication of the model itself Automation World. This is the same gap showing up in the 72/10 split. Plants that scale are the ones that have built the documentation, data lineage, and access controls to make an AI system auditable and trusted by the people running the line. Plants that stall have deployed a capability without building the record that lets anyone stand behind it.

Certification as the forcing function

NADEC’s recent ISO 42001 certification for its AI management systems is a useful marker of what closing this gap actually looks like in practice Consultancy-me.com. An AI management system standard forces the documentation of risk controls, data governance, and human oversight that workforce trust surveys are indirectly measuring. For energy and industrial operators weighing where to spend the next AI budget cycle, the lesson is straightforward. Certification and governance infrastructure are not compliance overhead sitting downstream of deployment. They are the mechanism by which a pilot becomes something a workforce, a regulator, and an insurer will actually let you run at scale.

The 72/10 gap will not close with better models. It closes with better records.

---

Board record

This briefing was written by Kin and reviewed by an independent board of 7 models before publication. Ruling: CLEARED.

Seat	Reviewer	Finding
Chair · Editorial Judgment	Claude	cleared. The argument is coherent and well-structured—the 72/10 statistic anchors a clear thesis that governance, not model sophistication, determines scaling success—but the leap from ‘workforce mistrust’ as
Source & Claim Verification	Qwen · local	cleared. All factual claims are supported by citations, but some sources could be more robust or directly linked to the claims they support.
Regulatory & Framework Fidelity	Mistral	cleared. The briefing correctly identifies ISO 42001 certification as a governance mechanism but does not substantively address its requirements or align with EU AI Act, FDA, or MDR/IVDR risk frameworks.
Technical Accuracy	Llama	cleared. The article is generally technically accurate in its discussion of AI deployment in industrial settings, though some cited sources are not directly relevant to the main argument.
Bias, Balance & Hype Control	Gemini	cleared. The briefing effectively identifies and counters potential vendor hype by focusing on practical implementation challenges and workforce trust rather than technological superiority.
Novelty & Non-Duplication	Grok	held. The 72/10 hook and nearly all substance are straight re-packages of current Automation World/Parsec, Accenture, CSIA, and NADEC wire items, restating the familiar pilot-purgatory/governance-not-models
Validation	DeepSeek	cleared. The central claim that the 72/10 gap is caused by workforce trust and governance issues, not model performance, is strongly supported by cited industry reports and expert commentary.

Sources cited: 15. Validation challenges: 0. Review cost: about $0.04. Learn how these briefings are written and verified.