RECENT EPISODES
Thu 20 AugVerified Data is the Missing Piece of Agentic Infrastructure With Gary Kotovets (Chief Data & Analytics Officer at Dun & Bradstreet)Gary Kotovets, Chief Data and Analytics Officer at Dun and Bradstreet, makes the case that verified data is the foundational requirement for agentic AI to function reliably, arguing that bad data and hallucination are mutually reinforcing dangers that compound as autonomous workflows reduce human oversight.Thu 6 AugGoverning AI That Keeps Evolving With Maryam Ashoori (VP of Product and Engineering at IBM watsonx.governance)Maryam Ashoori, VP of Product and Engineering at IBM watsonx.governance, draws on two decades of experience with multi-agent systems to argue that the core governance, risk, and compliance challenges have not fundamentally changed even as agentic AI has surged into enterprise production.Thu 23 JulWhy Smarter AI Agents Still Break With Juhi Parekh (GM of Key Frontier AGI Accounts at Turing)Juhi Parekh, GM of Key Frontier AGI Accounts at Turing, explains why a smarter model does not equal a more reliable product, arguing that agent deployment is fundamentally a system design problem rather than a research problem. She defines a well-constructed post-training task as one where pass-at-8 on a frontier model like Opus 4 or GPT-5 falls between zero and 30 percent, because ambiguous success conditions directly increase reward hacking.Wed 20 MayAI Agents Have an Identity Complex With Jeff Malnick (VP Engineering, Developer and AI at 1Password)Jeff Malnick, VP Engineering for Developer and AI at 1Password, explains why AI agents create a fundamentally new identity problem: unlike traditional machine workloads where permissions are delegated once at startup, agents decide mid-task that they need access to different systems, turning an already overprivileged baseline into what Malnick calls potentially infinite overprivilege.Wed 20 MayAI Agents Have an Identity Complex With Jeff MalnickIn this episode, Jeff Malnick explains why AI agents have turned a long-standing software identity problem into a near-crisis, arguing that because agents reason dynamically rather than executing fixed logic, they can demand access to entirely new systems mid-task, producing what he calls potentially infinite overprivileging on top of a baseline where 90 to 95 percent of permissions are already excessive.Thu 23 AprLessons from a Physician-CIO on AI Governance with Dr. Stacey Johnston (Chief Information and Digital Execution Officer at Beacon Health System)Dr. Stacey Johnston, a practicing hospitalist who became Chief Information and Digital Execution Officer at Beacon Health System in October 2024, discusses how she built AI governance from scratch at an organization that had no formal IT structure when she arrived, including an executive steering committee, eight advisory levels, and a dedicated AI council that requires a defined ROI before approving any initiative.Thu 23 AprLessons from a Physician-CIO on AI Governance with Dr. Stacey JohnstonDr. Stacey Johnston highlights the transformative potential of AI in healthcare, emphasizing its role in reducing physician burnout and streamlining workflows. She discusses the establishment of an AI governance framework to ensure responsible use and address privacy risks, particularly with large language models. Additionally, Johnston introduces innovative concepts like the "AI citizen program" for creating low-code solutions and stresses the importance of trust-building through successful implementation and engagement with clinical staff.Fri 10 AprThe Agentic Gap: What Enterprises Think vs. What Actually Works With Jeff Dalton (Head of AI and Chief Scientist at Valence)Jeff Dalton, Head of AI and Chief Scientist at Valence, joins to explain why most enterprise assumptions about agentic AI diverge sharply from what actually works in production.Fri 10 AprThe Agentic Gap: What Enterprises Think vs. What Actually Works With Jeff DaltonIn the podcast, Jeff Dalton discusses the "Agentic Gap," highlighting disparities between enterprise expectations and practical outcomes in AI applications. He emphasizes the importance of personalized coaching through Nadia, a transformative chat assistant, and advocates for precision in defining AI "agents" and evaluation metrics. The conversation also addresses the evolution of AI from deep learning to generative models, along with the challenges in measuring performance and integrating blockchain for secure data handling in financial services.Thu 5 FebHow to Prevent AI Agents from Going Rogue With David KennyDavid Kenney discusses the critical need for governance in AI to prevent agents from going rogue, emphasizing human oversight and real-time monitoring. He highlights the importance of collaboration across teams to establish trust and create standardized principles akin to GAAP for accountability in AI. Additionally, Kenney addresses the integration of AI in industries like finance, stressing the balance between safety controls and innovation, while also evaluating the cost implications and performance of AI systems.Thu, 25 Sept 2025Building Agents at Scale: Lessons from the Front Lines With Gary StaffordGary Stafford discusses the challenges of AI adoption in enterprises, emphasizing the need for targeted problem-solving rather than technology for its own sake, with a focus on agentic systems for complex decision-making. He highlights the importance of integrating AI with existing processes and frameworks to enhance operations, particularly in customer service and sales. Additionally, the conversation addresses the significance of multi-agent systems, which can improve task automation and communication within organizations while managing complexities related to orchestration and agent interactions.Sat, 16 Aug 2025Lessons Learned from Building Agentic Systems With Jayeeta PutatundaThe discussion with Jayeeta Putatunda on building agentic systems emphasizes several key themes: the importance of prioritizing high-impact AI use cases over extensive prototyping and the rapid obsolescence of AI frameworks. It highlights the need for rigorous evaluation and governance to manage biases in fully autonomous agents, particularly in sensitive areas like finance. The exploration of hybrid models combining traditional machine learning and LLMs aims to improve reliability while addressing the complexities in assessing agentic systems.Thu, 24 Jul 2025Agent Wars: The Hype, Hope, and Hidden Risks with Nate B. JonesIn the episode "Agent Wars: The Hype, Hope, and Hidden Risks," Josh Rubin and Nate B. Jones explore the complexities and challenges of AI agent adoption, emphasizing the need for strategic integration and a foundational understanding of data and operations. The discussion highlights the risks associated with multi-agent systems, the importance of clear constraints, and the potential for improving AI effectiveness through collaboration with domain experts. Additionally, they address the ongoing tensions between data privacy and accessibility, crucial for optimizing AI models.Thu, 20 Mar 2025AI Observability and Security for Agentic Workflows with Karthik BharathyThe discussion focuses on AI observability and security for agentic workflows, emphasizing the integration of robust governance, access controls, and monitoring to ensure data integrity and model performance. Karthik Bharathy highlights the evolution of generative AI, advocating for comprehensive observability tools and federated governance frameworks to balance standardization with flexibility. The conversation also addresses the challenges of security and data privacy in AI and blockchain applications, underlining the necessity for effective risk management and human oversight in automated systems.Thu, 27 Feb 2025GenAI Use Cases and Challenges in Healthcare with Dr. Girish NadkarniDr. Girish Nadkarni emphasizes the safe and ethical implementation of AI in healthcare, particularly its role in automating back-office tasks and enhancing clinical applications like risk prediction and sepsis detection. Generative AI's potential for personalized medicine and improved physician-patient interactions is notable, though challenges such as algorithm accuracy and ethical governance persist. Additionally, the episode discusses promising GenAI use cases like clinical summarization and diagnostic imaging advancements expected in the next 12-18 months.