45ms per frame, demonstrating near real-time responsiveness suitable for complex mining workflows. 2. Visual Resilience: The agent successfully located the UI elements even when the application window was moved to different coordinates, partially obscured, or resized, validating the robustness of the OpenCV module over traditional RPA tools. 5. DISCUSSION The separation of the "Cognitive Brain" (Orchestrator) from the "Haptic Body" (Local Agent) offers significant advantages for mining operations: 1. Security: The mining software never connects to the internet. The Local Agent acts as an airgapped bridge, receiving only sanitized commands from the internal Orchestrator. 2. Maintainability: If the mining software updates its interface (e.g., a button changes icon), only the reference image in the Local Agent needs updating. The complex logic in the Orchestrator remains untouched. 3. Scalability: A single Orchestrator can manage a fleet of agents across different engineering disciplines (Geology, Planning, Geotech). 6. CONCLUSION Mining Software Agents (MSA) represent a paradigm shift from static automation to Agentic AI. By combining the flexibility of LLMs with the safety of Symbolic Planning and the universality of Computer Vision, we provide a pathway to automate the "un-automatable" legacy systems in mining. The immediate benefits include the liberation of human capital from repetitive interface interactions and the reduction of manual data entry errors. Future development will focus on expanding the "Knowledge Base" to include complex geological modeling workflows, enabling engineers to act as supervisors of a digital workforce rather than operators of software. REFERENCES Xi, Z., et al. (2023). The Rise and Potential of Large Language Model Based Agents: A Survey. arXiv preprint. Fikes, R. E., & Nilsson, N. J. (1971). STRIPS: A new approach to the application of theorem proving to problem solving. Artificial Intelligence. Van der Aalst, W. M. (2018). Process Mining: Data Science in Action. Springer. Lacity, M., & Willcocks, L. (2016). Robotic Process Automation at Telefónica O2. The London School of Economics.
RkJQdWJsaXNoZXIy MTM0Mzk2