How We Use AI Safely 

At Velocity NDT™, artificial intelligence is used carefully, conservatively, and transparently. 
Our approach is designed specifically for regulated environments where accountability, traceability, and human judgment are essential. 

AI is a tool—not a decision maker. 

Our Philosophy on AI 

We believe AI should: 

  • Support qualified professionals, not replace them 

  • Reduce risk, not introduce new uncertainty 

  • Be auditor-friendly, not opaque 

  • Operate under documented controls, just like any other quality tool 

Every aspect of Aegis™ is designed around these principles. 

 

What AI Is Used For 

Within Velocity NDT™, AI is used only as a decision-support tool to: 

  • Compare procedures against defined regulatory or customer requirements 

  • Identify potential gaps, ambiguities, or missing information 

  • Summarize training, certification, and OJT records 

  • Highlight areas that require qualified human review 

AI helps surface issues earlier and more consistently—nothing more. 

 

What AI Is NOT Used For 

AI is explicitly prohibited from: 

  • Determining compliance or nonconformance 

  • Certifying personnel or approving training 

  • Approving procedures or revisions 

  • Replacing Level 3 authority or quality management judgment 

  • Acting as audit evidence without human verification 

All compliance decisions remain the responsibility of qualified human personnel. 

 

Human-in-the-Loop by Design 

Every AI-assisted output generated by Aegis™: 

  • Is reviewed by a qualified human 

  • Requires human approval before use 

  • Includes reviewer identification and date 

  • Is treated as decision-support information only 

No AI output is used directly in audits without human validation. 

 

Documented Governance & Controls 

Aegis™ operates under a formal AI governance framework, including: 

AI Use Policy 

Defines: 

  • Permitted and prohibited AI uses 

  • Human authority and accountability 

  • Disclosure requirements 

  • Data handling expectations 

AI Validation & Test Protocol 

Confirms AI fitness for its intended use only, including: 

  • Prompt adherence testing 

  • Traceability checks 

  • Hallucination resistance (negative testing) 

  • Repeatability verification 

  • Revalidation upon system or requirement changes 

These documents are available to clients and may be provided to auditors. 

 

Controlled & Predictable AI Behavior 

AI behavior within Aegis™ is constrained through: 

  • Locked, version-controlled prompts 

  • Explicit instructions to cite source text 

  • Requirements to flag missing or insufficient data 

  • Prohibitions against assumptions or conclusions 

This prevents “black-box” behavior and ensures predictable, reviewable outputs. 

 

Data Handling & Confidentiality 

  • AI uses only client-provided data 

  • Client data is not used to train public AI models 

  • Data handling follows contractual, regulatory, and confidentiality requirements 

  • No permanent storage of client data beyond agreed scope 

 

Transparency for Auditors 

Aegis™ is designed to be explained—not hidden—during audits. 

Clients receive: 

  • A one-page “AI for Auditors” explainer 

  • Clear documentation of AI limitations 

  • Validation evidence and governance controls 

Auditors see controlled analysis with human accountability, not automation replacing compliance. 

 

Why This Matters 

In regulated industries, unmanaged AI introduces risk. 

Aegis™ reduces risk by ensuring: 

  • AI is used conservatively 

  • Human authority is preserved 

  • Outputs are traceable and defensible 

  • Auditors are informed, not surprised 

 

Our Commitment 

We are committed to using AI responsibly, transparently, and in alignment with regulatory expectations. 

If AI cannot be used safely for a task, we do not use it.