Get your Data Ready for AI – or Let AI Deal with the Mess

AI promises smarter, more efficient buildings, but real-world deployment is rarely simple. The main barriers are often poor data quality, inconsistent naming, legacy systems, integration effort and limited operator buy-in. 

Drawing on experience at Monash University and other organisations, this presentation explores the value of structured building data using frameworks such as Brick, Haystack and VBIS. It will examine how better data can reduce onboarding time, improve portability between analytics platforms and support more effective use of tools such as fault detection, AI optimisation and general-purpose analytics. It will also cover the practical challenges of AI edge and cloud deployment, including BACnet and Modbus integration, cybersecurity, supervisory control and vendor-specific configurations. 

The presentation will explore when AI delivers enough value to justify the cost and complexity? For large, complex buildings and portfolios, it may. For many others, simple rules and better use of existing systems may still provide the best return. In the short term AI may prove most valuable in helping organise and validate the messy data holding buildings back.