Watson AI™ by Yehuda - CVD/HPHT Diamond Detector

The Yehuda Watson AI™ is a compact, rechargeable diamond screening system developed for professional identification and analysis of natural diamonds, CVD and HPHT lab-grown diamonds, CZ, moissanite and other simulants.

Combining Yehuda's established diamond detection technology with cloud-based AI-assisted classification, Watson AI provides an efficient solution for testing both loose stones and mounted jewellery.

The system can accommodate up to 8 rings simultaneously or approximately 50 carats of loose diamonds within its compact checking area.

AI analysis applies clear colour-coded markers to individual stones, helping the operator distinguish Natural Diamond, HPHT, CVD, CZ/other simulants and Moissanite. An uncertain AI classification is displayed as Refer, prompting the operator to reposition the stone, retest and use manual interpretation.

With rechargeable operation, compact dimensions, cloud result storage and a searchable Back Office, Watson AI is particularly suited to jewellers, diamond dealers, gemologists and other professionals requiring a portable diamond screening solution.

Key Features

• Portable professional diamond and lab-grown diamond screener
• Screens Natural, CVD and HPHT diamonds
• AI-assisted classification of different stone types
• Identifies CZ, Moissanite and other simulants
• Positive identification of Cubic Zirconia (CZ) using distinctive pink fluorescence
• Clear colour-coded AI stone classification
• Test as many as 8 rings simultaneously
• Test approximately 50 ct of loose diamonds
• Suitable for loose stones and mounted jewellery
• Checking area approximately 4.5 × 5 cm
• Approximately 4 cm checking height
• Rechargeable portable construction
• USB-C charging
• Cloud-based AI analysis
• Cloud storage for test records
• Searchable Back Office
• Multi-user management
• Store photographs, certificates, SKUs, names and other information with test records
• AI confidence information available through the Back Office
• One-year full warranty
  

Stones Screened & Classified
Watson AI is designed to assist professional users with the screening and classification of:

• Natural Diamonds
• CVD Lab-Grown Diamonds
• HPHT Lab-Grown Diamonds
• Cubic Zirconia (CZ)
• Moissanite
• Other Simulants
  

The AI component should be used as supporting information rather than as the sole or final determination of stone identity.

AI Colour Classification System
Watson AI uses a simple colour-based system to make AI analysis quick to interpret.

• Blue – Natural Diamond
• Red – HPHT
• Green – CVD
• Pink – CZ / Other Simulants
• Orange – Moissanite
• Black – Refer / AI Uncertain
  

When a stone is shown in black, the AI has not made a sufficiently certain classification. The stone should be repositioned and tested again, followed by manual interpretation where required.

All result screens are marked accordingly, and individual screens can be selected and enlarged for closer examination.

For reviewing results, Yehuda recommends viewing the original test image with the AI markings applied, although other available views can also be selected.

Positive Identification of Cubic Zirconia
Watson AI incorporates a specific method for recognising Cubic Zirconia (CZ).
CZ displays a distinctive pink fluorescence, providing the system with an additional characteristic for positive identification.
This helps the operator distinguish CZ when screening diamonds alongside other simulants.

Testing Capacity
The compact Watson AI chamber provides sufficient space for batch testing of both loose diamonds and mounted jewellery.

Mounted Jewellery
Test up to 8 rings at the same time, allowing multiple pieces to be screened within a single testing session.
Loose Diamonds
Approximately 50 carats of loose diamonds can be checked, making the system suitable for professional batch screening.
Testing Area
- Checking Area: Approx. 4.5 × 5 cm / 1.5 × 2.5"
- Checking Area Height: Approx. 4 cm / 1.5"

How to Activate the AI Function
Complete or save the test before switching the AI function to ON.
The system will then ask whether the stones being analysed are:
- Loose stones, or
- Mounted stones

Once selected, the cloud-based AI processes the test and applies the appropriate colour-coded markings to the stones.

If the test has been saved, the generated AI results are also stored.

Once processing has finished, the AI results can be reopened without waiting for the analysis to run again.

If a test is downloaded after the AI analysis is complete, an additional image containing the AI markings over the original stone image is included.

AI Confidence Levels
Detailed AI confidence information is available through the Watson AI Back Office.

After selecting a test and activating its AI analysis, users can enlarge the result image and view the confidence percentage assigned to individual stones.

A value approaching 100% indicates greater AI confidence in that particular classification.
Confidence information should still be considered alongside the operator's manual interpretation rather than being treated as a standalone identification.

Important Information About Watson AI
The Watson AI classification system is an advisory tool and not the final authority for determining stone identity.

Yehuda recommends using the system in the following way:
1. Manual interpretation should be the primary assessment.
2. Use the AI classification as a secondary check.
3. When the manual interpretation and AI result differ, reposition the stone and perform another test.
4. If uncertainty remains, use manual review and appropriate additional gemological testing.
5. Discrepancies can be reported to Yehuda Support to assist with ongoing AI development.
AI Development & Accuracy

The Watson AI feature is currently in an early and ongoing stage of development.
According to the manufacturer information supplied, internal testing has already demonstrated accuracy above 99%.

Yehuda's goal is to continue improving performance as the AI is exposed to larger numbers of stones. Their experts estimate that approximately 100,000 stones within each category will need to be tested for the AI to reach its intended optimal accuracy level.
Software development and updates are therefore ongoing.

Factors That Can Affect AI Results
Correct preparation and stone positioning are important when using AI analysis.

1. Dirty Stones
   Loose and mounted stones should be thoroughly cleaned before testing.
   Contamination can interfere with the image and potentially affect AI interpretation.
   
2. Overlapping Loose Stones
   Loose stones should be positioned individually without overlapping.
   
3. Rings with Side Stones
   For jewellery containing side stones, test the piece from multiple angles to obtain better coverage.
   
4. Strong Reflections
   If reflections interfere with the image, slightly reposition the stone or jewellery and repeat the test.
   
5. HPHT “Flashlight Effect”
   If an HPHT stone produces a strong glow, remove that stone and test nearby stones separately where necessary.
   
6. Stones Positioned in Corners
   Stones close to the corners of the testing area can be more difficult for the AI to determine.
   Move them closer to the centre of the tray and repeat the test.
   
7. Black / Unclassified Stones
   If the AI does not mark a stone or displays it in black, reposition the stone and perform another test.
   If the same result occurs again, use manual interpretation.
   
8. Tags or Paper
   Do not allow tags, labels or paper to remain visible within the checking area.
   These materials can reflect light and may cause the AI to generate an incorrect result.
   

Using AI Results Correctly

• Primary Assessment
  Always use the operator's manual reading first.
  
• Secondary AI Check
  The AI classification should then be used as additional supporting information.
  
• When Results Disagree
  If the AI classification differs from the manual interpretation:
  - Reposition the stone
  - Perform another test
  - Review the new result
  - Use manual assessment where required
  - Report relevant discrepancies to Yehuda Support
  

This feedback can also contribute to the continued improvement of the AI system.
Cloud-Based AI Processing

The AI function operates through the cloud and therefore requires an active internet connection.

Connection speed can affect the amount of time required to generate an AI result.
Importantly, the operator does not necessarily have to wait for processing to finish before continuing work. A test can be saved and submitted for AI analysis while additional stones are being tested.

Once an AI result has been generated and saved, it can be loaded immediately when that test is reopened.

Connection Information
The AI models do not incorporate Bluetooth or wireless communication.
The system operates using the specified connecting cable.

Although the AI analysis itself is cloud-based and requires internet access, this should not be confused with the detector having built-in Bluetooth or wireless communication.
Cloud Storage & Back Office

Watson AI provides cloud-based record management for storing and reviewing testing information.
Users can:
- Store previous test results
- Upload and save pictures
- Add certificates
- Record SKUs
- Add names and other identifying information
- Retrieve previous tests
- Compare previous results
- Delete unwanted records
- Move results to personal cloud storage
- Manage multiple users
- Review AI-generated results
- View individual AI confidence percentages

This provides a useful digital record for professional testing and inventory management.

Subscription & Storage
Use of the connected cloud services requires a subscription.
Standard Subscription
- Price: $12.50 per month
- Users: Unlimited
- Detectors: Up to 10 detectors per account
- Cloud Storage: Up to 1,000 test results
- Software Updates: Included

Additional storage can be purchased for an extra fee when larger volumes of test records need to be retained.

Results can also be deleted or moved to the user's own cloud storage to provide greater control over stored data.

Please note: the $12.50/month subscription should be considered when purchasing the Watson AI system.

Portable & Rechargeable Design
Watson AI is designed to provide professional screening capability without requiring a large permanent laboratory installation.

Its compact dimensions and rechargeable operation make it suitable for use in jewellery stores, diamond offices, laboratories, buying appointments and trade events.
- Dimensions: Approx. 90 × 140 × 102 mm
- Weight: Approx. 0.85 kg
- Charging: USB-C
- Design: Portable, lightweight and rechargeable

Specifications
- Brand: Yehuda
- Model: Watson AI™
- Product Type: Portable AI-Assisted Diamond Detector / Synthetic Diamond Screener
- Diamond Screening: Natural, CVD & HPHT
- Additional Classification: CZ, Moissanite & Other Simulants
- CZ Detection: Positive identification using distinctive pink fluorescence
- AI System: Cloud-based, colour-coded stone classification
- Mounted Jewellery Capacity: Up to 8 rings simultaneously
- Loose Diamond Capacity: Approx. 50 ct
- Checking Area: Approx. 4.5 × 5 cm / 1.5 × 2.5"
- Checking Area Height: Approx. 4 cm / 1.5"
- Power: Rechargeable
- Charging Connection: USB-C
- Dimensions: Approx. 90 × 140 × 102 mm
- Weight: Approx. 0.85 kg
- Internet: Required for cloud-based AI analysis
- Bluetooth / Wireless Communication: No – AI models operate through the specified connecting cable
- Cloud Storage: Up to approx. 1,000 test results with standard subscription
- Users Per Account: Unlimited
- Detectors Per Account: Up to 10
- Subscription: $12.50/month
- Software Updates: Included with subscription
- Additional Storage: Available for an additional fee
- Back Office: Yes
- AI Confidence Levels: Available through Back Office
- Warranty: 1-year full warranty

Why Choose the Yehuda Watson AI™?
Watson AI combines portable diamond screening, AI-assisted classification and batch testing in one compact rechargeable system. Test approximately 50 ct of loose diamonds or up to 8 rings at once, with cloud-based storage for easy result management.

Important Notes
AI results are advisory and should support manual interpretation, not replace it. Uncertain results should be repositioned and retested. An active internet connection is required for AI analysis, and the AI continues to improve through software updates.