This session presents an architecture for 'knowledge On-demand' environments, along with an explanation of how government is publishing machine-readable Linked Data to support knowledge inferencing and semantic interoperability. Spatial information applications for climate change, mining and agriculture will demonstrate the potential of machine learning for on-demand evidence-based decision-making and progress towards achieving Sustainable Development Goals. With the explosion of data from a variety of sources, spatial data infrastructures need to evolve from simply being a distribution channel for data; to one that allows people to interrogate data using natural language queries to gain new information. This 'leapfrog' technology has potential to accelerate the value of geospatial information. Advances in BIG data, distributed processing, artificial intelligence, Semantic Web technologies and the intensive use of mobile devices, have dramatically increased the knowledge value of geospatial data and the appetite of consumers for real time information.