This talk will demonstrate how the application of large language models can help to realise value from well managed information through an application we call ProjectGPT. The volume of information generated on large infrastructure projects like the M5 Metro in Copenhagen is significant. Good information management practices are essential to create a reliable and trustworthy "single source of truth", however when the source(s) of truth becomes too large or complex, people cannot find the information they require for their job in any reasonable timeframe. ProjectGPT solves this problem by replacing traditional keyword search functionality with a Retrieval Augmented Generation chat bot, which enables project members to ask questions about the project or topic of interest and only receive responses specific to the project
Speaker:
Rory McCully
Associate Director, Arup
