AI Usage & Transparency

1. AI in the product

1.1 NDT Image Analysis USES AI

The NDT module of ENODEUS uses a YOLOv8-based computer vision model to assist certified NDT inspectors in analyzing liquid-penetrant (PT) inspection images. This model is an assistive tool only: it does not replace the professional judgment of a certified inspector and does not make accept/reject decisions. See the Terms of Service, Sections 2 and 4, for the full description and limitations.

1.2 NDT Asset Traceability NO AI

The NDT Asset Traceability module — including asset management, QR code tracking, inspection history, Non-Conformance Reports (NCRs), Corrective and Preventive Actions (CAPAs), and audit trace logs — does not use any AI or machine learning model. All data in this module is entered, reviewed, and validated by human inspectors and managers. The immutable trace log is a deterministic append-only record, not AI-generated.

1.3 Other modules NO AI

All other modules of ENODEUS (Welding, Calibration, Electrical, Construction, Offshore, Hospitality, Audits, Transport, and others) do not use any AI or machine learning model. Reports generated by these modules are based entirely on data manually entered by the user.

2. AI in the development of this app

Parts of ENODEUS's software (source code, configuration, and supporting scripts) were developed with the assistance of artificial intelligence tools, used as a coding aid under the direction, review, and final decision-making of the founder and sole developer. Every architectural choice, business rule, data model, security policy, and feature decision in ENODEUS was made by the founder; AI tools were used to accelerate implementation, not to originate product or business decisions.

3. Development tools and workflow

ENODEUS is built using an industry-standard software development workflow. Tools used in the development process include:

• Xcode 26 — primary IDE for building, compiling, archiving, and submitting the iOS application.
• Swift & SwiftUI — the programming language and UI framework the app is written in.
• Terminal (zsh) — direct command-line control of the build, scripting, and deployment environment.
• Python — used for development-time automation scripts (file edits, validation, security testing); not part of the shipped application.
• Roboflow — used locally during model development to prepare and organize the image dataset used to train the YOLOv8 computer vision model. Not part of the shipped application and not connected to images uploaded by users in production.
• Git & GitHub — version control and source code history for the entire codebase.
• Firebase CLI — deployment of Firestore security rules, composite indexes, and backend configuration.
• Firebase / Firestore / Auth / Crashlytics / App Check — backend infrastructure, database, authentication, crash reporting, and API security.
• App Store Connect — distribution, TestFlight beta testing, and App Store submission.
• Stripe (planned) — payment processing for Company subscriptions. Not yet active in production.
• Claude (Anthropic) — the primary AI coding assistant used throughout development, for writing, reviewing, and debugging code under direct human supervision and direction at every step. All architectural and product decisions remained with the founder; Claude was used as an implementation aid, not as an autonomous decision-maker.
• Document and note-taking tools — used for internal project documentation and planning notes.

4. Human oversight and decision-making

All product decisions — including data architecture, user roles and permissions (Inspector, Manager, Company), business logic, security policies, pricing, and legal terms — are made and approved by the founder before implementation. AI-assisted code is reviewed, tested, and compiled by the founder before being committed to the production codebase. Firestore security rules are manually audited and tested before each deployment. ENODEUS maintains an internal development log documenting significant decisions and their rationale.

5. Security and access control

No AI is used in ENODEUS's security or access control systems. Role-based access control (Inspector, Manager, Company) is implemented through deterministic Firestore security rules, manually written and audited. Firebase App Check is used to verify that requests originate from the legitimate app binary.

6. Questions

For questions about this disclosure, contact: contact@enodeus.ch