Hexagon’s HxGN Alix: AI Assistant for Asset Management
Hexagon’s Asset Lifecycle Intelligence division, in collaboration with AWS, has developed HxGN Alix, an AI-powered digital worker designed to revolutionize user interaction with Enterprise Asset Management (EAM) products. This AI assistant aims to streamline information access, enhancing internal workflows and improving customer satisfaction by providing quick, accurate answers to user queries, eliminating the need to navigate extensive PDF manuals. HxGN Alix is built to operate securely within high-security environments, addressing specific challenges faced by clients while upholding stringent security standards. The development strategy employed a phased approach: ‘crawl’ (establishing secure infrastructure), ‘walk’ (integrating customer data), and ‘run’ (developing high-value use cases). The technology stack leverages several AWS services including Amazon EKS for compute and model deployment, Amazon EC2 G6e instances with NVIDIA L40S GPUs for cost-effective generative AI model hosting, and Mistral NeMo, a 12-billion parameter open-source LLM for multilingual capabilities and function calling. Amazon Bedrock acts as a fallback for regional failures, while Amazon Bedrock Guardrails ensures safety and compliance. Amazon S3 provides secure data storage, and Amazon Bedrock Knowledge Bases enables Retrieval Augmented Generation (RAG) for improved response accuracy. The development journey included addressing challenges like maintaining context during document chunking, handling diverse document formats, managing LLM boundaries, grounding responses, and handling brief, context-dependent follow-up questions. Two approaches were tested for handling one-line conversation follow-ups: prompt-based search reformulation and context-based retrieval with chat history. The system also needed adjustments to the traditional application development lifecycle to manage the inherent variability and testing challenges of generative AI systems. Overall, HxGN Alix showcases the transformative potential of generative AI in enterprise asset management, offering significant improvements in efficiency and user experience within secure environments.
HxGN Alix represents a significant advancement in ai automation asset management, streamlining workflows and enhancing decision-making for industrial operations.
While many companies explore chatgpt automation ai solutions for various applications, Hexagon’s HxGN Alix specifically targets industrial asset management workflows.
