Agentic AI, which has swiftly transitioned from a novelty to a core business strategy, is still a concept many organizations struggle to integrate into daily operations, as discussed in the "Agents of Data" podcast featuring experts from Matillion. The conversation highlights the challenges organizations face in becoming agentic AI-ready, emphasizing the need for both data and personnel preparedness, with issues like fragmented ownership and poor documentation impeding progress. Matillion's own engineering teams have embraced a spectrum of agentic tool usage, encouraging experimentation and identifying super-users to champion the technology. The podcast further explores extending agentic workflows beyond engineering, advocating for baseline AI training and focusing on automating mundane tasks. The theme of constraint is prominent, with safe agentic systems requiring defined problem spaces and human-in-the-loop processes for high-stakes actions. Matillion's innovative "gym" framework for testing agentic systems under varying conditions is presented as a solution for adapting traditional binary testing methods. The discussion also addresses the potential of agentic AI in processing unstructured data, the importance of governance, and the risks of shadow AI. The future of agentic AI is envisioned as a progressive augmentation of human work, likened to the evolution of smartphones, with Matillion's agentic data team, Maia, exemplifying the practical application of these concepts by automating data engineering tasks while maintaining governance and quality.