AI hazard classification

AI hazard classification

Verdan uses artificial intelligence to automatically classify safety observations, saving time and ensuring consistent categorization.

How It Works

When you submit a safety observation:

• Text Analysis — AI reads your description to understand the hazard
• Image Analysis — If photos are attached, AI identifies visual hazards
• Classification — The hazard is categorized into standard types
• Severity Assessment — Risk level is estimated
• Standards Reference — Relevant OSHA regulations are identified
• Recommendations — Corrective actions are suggested

Hazard Categories

The AI classifies into OSHA's Focus Four and other categories:

Focus Four (Most Common Fatalities)

• Falls — Unprotected edges, holes, scaffolding
• Struck-By — Falling objects, vehicles, equipment
• Caught-In/Between — Machinery, collapsing structures
• Electrocution — Exposed wiring, power lines, equipment

Additional Categories

• Fire/Explosion
• Chemical exposure
• Ergonomic hazards
• Environmental conditions
• Housekeeping
• PPE violations
• Traffic/vehicle
• Noise exposure

Severity Levels

OSHA References

The AI maps observations to relevant OSHA standards:

• 29 CFR 1926 (Construction)
• Specific subparts and sections
• Direct links to regulation text

Accuracy & Overrides

• AI classification accuracy: ~92% in testing
• Always review AI suggestions before accepting
• You can modify the category, severity, or recommendations
• Corrections help improve future accuracy
• Human judgment always takes precedence

Privacy & Data

• Observation data is processed securely
• Photos are analyzed in-memory, not stored externally
• AI models are updated quarterly
• No personally identifiable information is used in training