10 Internal Inconsistencies in 3 Published Groundwater Surveys

Developer Dann Waneri has built an AI-powered agent designed to assist geophysicists in verifying groundwater survey data. The project, submitted for the Sanity Challenge, addresses the critical issue of data integrity in borehole siting within the Niger Delta. By structuring data from three published geophysics papers into a Sanity dataset, the agent allows users to query resistivity readings and verify them against published figures. During development, the author identified 10 internal inconsistencies—such as swapped table labels, contradictory curve types, and erroneous depth arithmetic—that were present in the original research papers. The agent provides a multi-modal interface, including an offline rule-based mode that functions without network access or LLM dependencies. By combining structured data queries with automated code-based validation, the tool ensures that borehole recommendations are grounded in verified, consistent data rather than potentially flawed source material.