TADI: Tool-Augmented Drilling Intelligence via Agentic LLM Orchestration over Heterogeneous Wellsite Data

Rong Lu introduces TADI (Tool-Augmented Drilling Intelligence), an agentic AI system designed to transform heterogeneous wellsite data into analytical intelligence. Applied to the Equinor Volve Field dataset, TADI integrates 1,759 daily drilling reports (DDRs), selected WITSML real-time objects, 15,634 production records, formation tops, and perforations. The system uses a dual-store architecture: DuckDB for structured queries over 12 tables with 65,447 rows, and ChromaDB for semantic search over 36,709 embedded documents. Twelve domain-specialized tools, orchestrated by a large language model via iterative function calling, support multi-step evidence gathering that cross-references structured drilling measurements with daily report narratives. The entire implementation is framework-free, totaling 6,084 lines of code, and is reproducible using the public Volve dataset and an API key.

Key technical achievements include parsing all 1,759 DDR XML files with zero errors, handling three incompatible well naming conventions, and being backed by 95 automated tests plus a 130-question stress taxonomy spanning six operational categories. The paper formalizes the agent's behavior as a sequential tool-selection problem and introduces the Evidence Grounding Score (EGS) as a grounding-compliance proxy. Case studies and qualitative ablation analysis suggest that domain-specialized tool design, rather than model scale alone, is the primary driver of analytical quality in technical operations. This work demonstrates how agentic LLMs can be effectively specialized for highly technical domains like drilling operations.