TADI: AI System for Drilling Data Analysis
TADI, which stands for Tool-Augmented Drilling Intelligence, is an AI created to transform drilling data into meaningful insights. It works with the Equinor Volve Field dataset, which includes 1,759 daily drilling reports and various real-time data objects. The system employs a dual-store setup: DuckDB handles structured queries across 12 tables with 65,447 rows, while ChromaDB manages semantic searches through 36,709 embedded documents. A large language model orchestrates twelve specialized tools, integrating structured drilling info with daily report insights. TADI efficiently processes all 1,759 DDR XML files, adapts to three different well naming styles, and is backed by 95 automated tests and a comprehensive 130-question stress taxonomy across six operational categories.
Key facts
- TADI stands for Tool-Augmented Drilling Intelligence
- Applied to Equinor Volve Field dataset
- Integrates 1,759 daily drilling reports
- Uses DuckDB for structured queries and ChromaDB for semantic search
- 12 domain-specialized tools orchestrated by LLM
- Parses all 1,759 DDR XML files with zero errors
- Handles three incompatible well naming conventions
- Backed by 95 automated tests and 130-question stress taxonomy
Entities
Institutions
- Equinor
Locations
- Volve Field