Rio Tinto AI Documents Metpro System: Streamlining Legacy Tech

Rio Tinto has developed an AI assistant to document its 30-year-old Metpro manufacturing system. This system is crucial for their aluminum operations in Australia and New Zealand. The company's data science senior advisor, Ke Shi, explained that Metpro's documentation was fragmented across thousands of documents. This made onboarding new staff difficult and increased the risk of errors. To address this, Rio Tinto fed Metpro's code and business context into Amazon Bedrock Knowledge Bases and AgentCore. They then used Amazon SageMaker AI Jumpstart with Llama 3.1 8B as the initial AI model. This project helps centralize fragmented information and makes system behavior more discoverable without replacing the existing legacy platform. This approach is important for companies looking to integrate AI with their older operational technologies.