How AllShores lead a 6-week AI Proof of Concept for large Australian mining company
This case study showcases how AllShores, as a consulting partner with Davidson and AIRis, came together to transform large Australian mining company's mining studies process through innovative AI implementation. By developing a dual-solution approach with AI Agents and an AI Workbench, the team successfully reduced SEP (Study Evaluation Process) report creation from months to weeks while maintaining quality and establishing a foundation for future enterprise-scale deployment.
Delivered the complete proof of concept one full week ahead of the scheduled timeline, demonstrating exceptional efficiency.
Successfully ingested and analysed hundreds of complex technical documents, breaking them into thousands of manageable chunks for AI processing.
AI agents consistently produced SEP drafts at 70-80% completion level, requiring only final human refinement rather than starting from scratch.
Each iteration of prompt tuning delivered approximately 10% quality uplift, demonstrating continuous improvement capabilities.
large Australian mining company's Study Evaluation Process (SEP) reports were creating significant operational bottlenecks, with each report requiring between 3 to 9 months to complete. This lengthy process consumed valuable Subject Matter Expert (SME) bandwidth that could have been directed toward more strategic initiatives.
Critical knowledge was buried across hundreds of PDF documents, complex tables, and technical images, making information retrieval time-consuming and inefficient. The siloed nature of these documents created collaboration gaps that frequently led to duplication of efforts and missed insights that could impact project outcomes.
Importantly, large Australian mining company's executive team had a clear vision: they wanted AI capabilities embedded within their existing workflows rather than implementing a disconnected experimental solution. This required a thoughtful approach that would integrate seamlessly with their established processes.
We developed a comprehensive solution comprising two complementary components:
Specialized AI agents that could summarise prior studies, identify potential risks, and draft SEP sections autonomously. These agents functioned as virtual team members, augmenting human capabilities rather than replacing them.
A robust foundation featuring reusable pipelines and rapid agent cloning capabilities. The platform leveraged secure AWS Bedrock integration with OpenSearch deployment, ensuring enterprise-grade security whilst maintaining high performance.
This dual approach established a strategic platform for reuse across multiple projects, creating long-term value beyond the initial implementation and setting the foundation for enterprise-scale deployment.
Initial AWS limitations regarding SSL, region availability, and reasoning models were successfully resolved through collaboration with BDA, highlighting the importance of flexible technical approaches when implementing enterprise AI solutions.
SME oversight proved essential throughout the process. While AI drafts consistently reached 70-80% completion, human expertise remained critical for refinement, validation, and contextual understanding that AI currently cannot replicate.
The Workbench approach established a strategic platform for reuse across multiple projects, creating long-term value beyond the initial implementation and setting the foundation for enterprise-scale deployment.