DrawingPhin Engineering Data & AI Policy

DrawingPhin is built for mechanical drawings, FAIR preparation, and AI-assisted drawing review. We understand that engineering drawings are not ordinary documents. They may contain design intent, manufacturing requirements, quality requirements, customer information, and confidential engineering know-how.

This policy explains how DrawingPhin handles Engineering Data when you upload drawings, generate FAIRs, run drawing reviews, or export results.

1. What We Mean by Engineering Data

"Engineering Data" means the technical data you upload to or generate through DrawingPhin, including:

• mechanical drawings, PDFs, images, and rendered drawing pages;
• OCR text, text layers, word boxes, and extracted drawing text;
• dimensions, tolerances, GD&T, datums, notes, tables, and title block data;
• FAIR characteristics, balloon numbers, inspection requirements, and review results;
• user corrections, confirmations, edits, comments, and export settings;
• custom review rules and check criteria you author, including their titles and descriptions;
• exported files — FAIR forms as Excel (.xlsx), drawing review reports as PDF or Word (.docx);
• temporary processing artifacts such as crops, thumbnails, tiles, and intermediate OCR inputs.

Engineering Data is handled separately from Personal Data. Personal Data includes information such as your name, email address, billing details, IP address, device information, and marketing preferences. Personal Data is covered by our Privacy Policy.

2. Your Drawings Stay Yours

You retain ownership of your drawings, engineering files, FAIR data, review results, user corrections, and exported files.

DrawingPhin does not claim ownership of your Engineering Data. You grant DrawingPhin a limited right to process Engineering Data only as needed to provide the service you request, such as drawing analysis, FAIR generation, auto-ballooning, review findings, project history, export generation, troubleshooting, and deletion workflows.

3. How We Use Engineering Data

DrawingPhin uses Engineering Data to provide and improve the service, including to:

• read and process uploaded drawings;
• detect title block information and drawing metadata;
• extract dimensions, tolerances, GD&T, datums, notes, tables, and manufacturing requirements;
• generate FAIR characteristics and structured drawing data;
• create ballooned drawings and export files;
• generate drawing review findings, confidence scores, and source evidence;
• allow users to review, edit, confirm, and export results;
• maintain project history where supported by the user's plan;
• calculate usage, credits, processing cost, and system performance;
• troubleshoot errors, improve reliability, and protect the service from abuse.

DrawingPhin does not sell, rent, or transfer customer Engineering Data to third parties for their own commercial use.

4. AI and OCR Processing

DrawingPhin uses OCR, vision models, language models, rule-based parsers, and internal drawing intelligence systems to analyze engineering drawings.

To provide the service, Engineering Data may be processed by third-party providers that support hosting, storage, OCR, AI processing, authentication, payments, email, analytics, monitoring, or security. Current providers may include cloud infrastructure providers, OCR providers, and AI model providers.

We may add, remove, or replace providers, models, routing methods, or processing systems without individual notice, provided that we continue to apply this policy and reasonable data protection safeguards.

Major provider categories and subprocessors may be described in our Security or Subprocessor documentation.

5. No AI Model Training on Customer Drawings

DrawingPhin does not use customer drawings or Engineering Data to train AI models.

This includes uploaded drawings, rendered drawing images, OCR text, extracted features, FAIR characteristics, drawing review findings, user corrections, and exported files.

We also do not authorize third-party AI or OCR providers to use customer Engineering Data to train their models, unless the customer separately and explicitly agrees.

6. Product Improvement

DrawingPhin may use aggregated, de-identified, or non-content usage information to improve the service. Examples include:

• processing time;
• error rates;
• page counts;
• file type statistics;
• feature usage;
• model routing performance;
• credit usage;
• system reliability metrics;
• common failure patterns that do not reveal customer drawing content.

We may also use customer feedback, support requests, and technical diagnostics to improve DrawingPhin.

We do not publish customer drawings, show customer drawings in demos, use customer files as marketing examples, or build public datasets from customer Engineering Data without permission.

7. Human Access

DrawingPhin processes drawings through software systems, and human access to customer Engineering Data is limited.

The internal console our team uses to operate the service shows metadata only — file name, part number, processing status, token usage, and cost. It does not display drawing images, extracted characteristics, review findings, or FAIR content.

Beyond that console, support or engineering personnel access Engineering Data only when reasonably necessary to provide support, investigate a technical issue, address abuse or security concerns, comply with legal obligations, or maintain the service — and only the data needed for that task.

8. Data Retention and Deletion

**What you don't save, we don't keep.** Retention depends on one thing: whether an analysis was saved to your parts library.

**Saved to your parts library.** The drawing and its analysis results — characteristics, FAIR data, drawing review findings — are kept for as long as you keep them, so you can reopen, edit, export, and compare revisions from any device. They are removed when you delete them, or when your account data is removed (see below). Saving to the parts library is a subscription feature; it is not included in the Pay-as-you-go plan.

**Not saved to your parts library.** Analysis requires the drawing to reach our servers, so a copy is stored while the work runs. That copy is deleted automatically 24 hours after the analysis, together with the key material that opens it *and* the analysis results themselves. A scheduled job runs daily to carry this out. Nothing about the drawing remains on our servers afterwards — only the billing record described below.

**Billing records** — job identifier, timestamp, amount charged, token usage, and file name — are kept regardless, because they have independent accounting and dispute-resolution retention obligations. They contain no drawing content.

Analyses that were not saved to your parts library also remain in your browser's local staging area on the device that ran them, up to 10 at a time; older ones are pushed out and cannot be recovered. Users may delete them from their own device at any time.

If you cancel your subscription, you can delete your files first. Anything still remaining is removed within 30 days of cancellation.

Deleting saved project history from inside the product is not available yet. Until it is, contact us and DrawingPhin will remove the Engineering Data you identify from active systems, subject to backups, security logs, legal requirements, billing records, dispute resolution, and operational retention limits.

9. Export-Controlled and Restricted Data

DrawingPhin's standard version is not designed for ITAR, CUI, FedRAMP, government-controlled data, or other specially restricted data unless we separately agree in writing.

Users are responsible for ensuring they have the legal and contractual right to upload and process any drawing or technical data submitted to DrawingPhin.

Do not upload drawings or technical data if doing so would violate a customer contract, NDA, export control rule, supplier agreement, or applicable law.

10. AI Output Requires Review

DrawingPhin is an AI-assisted tool. It does not replace professional engineering judgment, quality engineering review, customer approval, inspection planning approval, or regulatory compliance review.

AI-generated results may be incomplete, inaccurate, or unsuitable for a specific customer requirement. Users are responsible for reviewing and validating all outputs before relying on them, including:

• characteristics;
• tolerances;
• GD&T callouts;
• balloon numbers;
• drawing review findings;
• source evidence;
• export fields;
• customer template requirements.

11. Changes to This Policy

We may update this policy as DrawingPhin evolves. Material changes may be posted on our website or communicated through the product where appropriate.

12. Contact

For questions about Engineering Data or AI processing, contact:

info@drawingphin.com