The Challenge
A construction materials company with a large trucking fleet operated in a reactive safety culture. Safety data was fragmented across multiple systems with no way to assess which upcoming deliveries posed higher risk. Safety manuals and incident reports were trapped in PDFs with no searchability.
Our Solution
We built a comprehensive safety intelligence platform on Snowflake with three core capabilities:
1. Unified Data Consolidation
Integrated 6 data sources into single Snowflake platform: fleet maintenance, telemetry, safety manuals, incident reports, delivery tickets, orders, and weather data.
2. Job-Level ML Risk Scoring
Deployed Snowflake ML pipelines to score deliveries/orders based on weather risk, seasonality, fleet age, team metrics, capacity constraints, and operational busyness—enabling proactive interventions before job assignment.
3. Natural Language Query Interface
Combined Snowflake Cortex Analyst (for structured data queries) and Cortex Search (for document intelligence) in single chat interface. Safety team can ask questions without SQL expertise.
Results
Investigation Time Reduction
Through unified data and natural language queries
Data Sources Unified
Single platform for all safety intelligence
Risk Assessment
Identify high-risk jobs before assignment
Key Impact
- Safety team shifted from reactive incident response to proactive job-level risk assessment
- Document search revealed patterns invisible when knowledge trapped in PDFs
- Natural language interface democratized insights—non-technical users can explore data
- Platform became foundation for data-driven safety culture vs. institutional knowledge