There is interest from academia and industry to investigate the application of Artificial Intelligence (AI)/Machine Learning (ML) to various use cases associated with the Air Interface of cellular systems, e.g., for reporting Channel State Information (CSI) feedback, for beam management, and for positioning accuracy. In this paper, we develop a research platform capable of real-time inference using an AI-enabled CSI feedback that closely represents real-world deployment scenarios. In our experiment, we evaluate the performance of the proposed framework by integrating a CSI autoencoder into the OpenAir-Interface (OAI) SG protocol stack. Further, we demonstrate the real-time functionality of the CSI compression framework with the encoder deployed at the User Equipment (UE) and CSI reconstruction with the decoder deployed at the Next Generation Node Base (gNB). The experiments are conducted on an Over-the-Air (OTA) indoor testbed platform, ARENA, as well as, on an emulated environment using Colosseum, the world's largest wireless network emulator.