Digitalization of Industrial Risk Analysis Using a Deterministic AI-Augmented Engine: A Field Engineer's Approach

Abstract—Industrial physical-security and site-survey risk analyses
still rely heavily on manual procedures: spreadsheets, personal
interpretation of ISO/IEC-aligned practice (notably ISO 31000 /
ISO/IEC 27005 frames and IEC/EN 62676-4 for CCTV), and narrative
reports that are slow, expensive, and difficult to audit. Two equally
experienced engineers may produce inconsistent recommendations
from the same facts, and economic justification (CAPEX, OPEX, ROI)
often drifts from dimensioned bill-of-materials evidence. This
technical note documents how a field electronics engineer with more
than forty years of experience—age seventy-two at the time of the
work, without formal programming or AI education—specified and
steered a Python-based prototype, RAP/RAE (Risk Analysis Platform /
Risk Analysis Engine). The engine digitizes an end-to-end
risk-to-investment workflow as sequential modules M0–M8,
producing auditable artefacts (BOM, CAPEX/OPEX, scenario
ranking, and HOLD / INVEST / CONDITIONAL dictamina). The
design principle is: AI does not decide the risk. The engine decides. AI
translates and explains. LLM assistance is confined to edges; risk
scoring, control recommendation, technical dimensioning, and
economic gates remain deterministic. The contribution shows that a
senior domain expert can digitize a complex engineering procedure by
commissioning AI with precise specifications, while deterministic
governance keeps AI from substituting for engineering judgment.
Index Terms—AI augmentation, deterministic AI, field engineering,
industrial safety, non-programmer AI use, prompt engineering,
RAP/RAE, risk analysis, senior engineering, Python automation