Key results
The company
Pilot Company
pilotcompany.comOperator of travel centers, truck stops, and gas stations across North America.
Result highlights
- Low 7-figure value via GenAI
- 90% processing cost reduction
- 99% reduction in manual processing
- 98.6% accuracy vs 71% human baseline
The story
North America’s largest travel center operator manages a fuel distribution network of 850 locations, delivering fuel every 22 seconds across the U.S. and Canada.
Drivers submit cell phone photos of shipping receipts that vary across 300 layouts with mixed handwriting and typed text. Traditional OCR tools failed to interpret these complex images, creating manual data entry bottlenecks that delayed revenue recognition.
The team implemented a multi-modal GenAI pipeline that processes raw document images to preserve spatial context like checkbox layouts. A validation layer uses fuzzy matching against internal reference tables to correct address discrepancies and verify product data. The system connects to Amazon Bedrock models through a unified gateway to standardize extraction across the diverse document types.
Quotes
“Our GenAI app, supported by Databricks, is delivering value in the low seven figures through a combination of cost savings and increased revenue.”