Madalina Florean, Director of AI at Cadmatic, examines why a digital twin can contain a wealth of project information yet still fail to earn users’ trust. If someone finds an error they can check for themselves, they may be reluctant to rely on the twin for answers they cannot verify. Her presentation focuses on finding and correcting those problems before they carry through to construction and operations.
A recurring example is an item of equipment given different tags by engineering, procurement and construction teams. Each team may be referring to the same pump, but the inconsistent names break the link between their records. Madalina explains how an early data correction can become a costly revision or delay when the mismatch is discovered later. She argues for connecting information across disciplines and checking its completeness as a project develops.
AI can help find mismatched tags in models, point clouds and older documents, as well as maintain the digital twin as information changes. Madalina also describes how it could help users ask questions across disciplines and see the documents, equipment and project status connected to an asset. Those answers must respect the same access permissions that apply outside an AI assistant.
For Madalina, trust depends on combining AI with checks that produce the same verdict when given the same data. Teams need to be able to review how a result was reached, audit changes and know who is responsible when information crosses between companies. These foundations make it possible to use AI to explore design options and support decisions without allowing hidden data errors to spread.
Watch the full presentation to see Madalina’s examples and demonstration of how connected information can help teams find answers in a digital twin.
Speaker
- Madalina Florean, Director of AI, Cadmatic
