This talk will cover the overview of Xilinx's new unified AI platform ranging from cloud to edge. It will cover the basic tool flow through pruner, quantizer, compiler, and run-time to achieve AI inference directly from frameworks like TensorFlow.