AI Confidential

Hosts Aaron Fulkerson and Mark Hinkle tackle the provocative question of whether large language models have hit a wall. With critics calling OpenAI's ChatGPT-5 'overhyped and underwhelming,' they discuss the slowdown in LLM advancements, why 95% of AI projects are failing, and how enterprises can harness AI without sacrificing data security. The episode explores the trough of disillusionment and what comes next for enterprise AI.

Kellie Romack knows deploying secure agents needs to be a baseline for enterprises. Since partnering with OPAQUE, Microsoft Azure, and NVIDIA to deploy confidential AI agents, ServiceNow has seen average response times from its sales commission help desk speed up from four days to just eight seconds.

Marco Palladino is on the forefront of integrating APIs with AI — and is wrestling with the questions around security and governance that this shift presents. In this episode, we talk about LLM gateways, how AI is impacting API dev, and how enterprises can prepare for the agentic web.

From Opaque's annual Confidential Computing Summit, hosts Aaron Fulkerson and Mark Hinkle interview key speakers about the agentic web, data security, and AI. Jason Clinton discusses the future of agents, MCP, and Anthropic's ASL-3 safety upgrade. Daniel Rohrer shares how NVIDIA handles AI security and scaling compute power. Daniel J. Beutel explains how Flower Labs uses federated learning to keep data secure in AI workloads.

Hosts Aaron Fulkerson and Mark Hinkle interview speakers at Opaque's Confidential Computing Summit. Mark Russinovich discusses new confidential AI developments from Microsoft Azure, including a framework for security levels in confidential computing. James Kaplan shares how enterprises are mining unstructured data with GenAI. Vinay Pillai explains how ServiceNow used confidential agents to cut sales commission help desk response times from four days to eight seconds.

Vijoy Pandey understands that an internet populated by agents requires an open, interoperable framework for agent-to-agent communication and a seismic shift in governance. That's why Outshift by Cisco, in partnership with LangChain and Galileo, created AGNTCY.

John Willis discusses the history of AI and the risks associated with data exhaust. Willis recently published Rebels of Reason, an AI history told through the stories of lesser-known technologists who built the foundation for modern-day AI.

Hosts Aaron Fulkerson and Mark Hinkle talk to Atin Sanyal about the growing reliability challenges in GenAI. The episode covers Galileo's ChainPoll hallucination detection methodology, which uses consensus scoring to improve output reliability. They discuss evaluation agents that get smarter over time, how enterprises can evolve their own AI quality metrics, and why data quality and confidential computing are becoming increasingly critical.

Hosts Aaron Fulkerson and Mark Hinkle talk to João Moura about building CrewAI, the agentic AI framework designed to make agent creation feel borderline magical. Inspired by the Ruby on Rails philosophy of convention over configuration, CrewAI prioritizes speed, clarity, and ease of use. The episode covers why interoperability across your stack is essential for scaling AI, how managers are uniquely equipped to build effective agents, and the dangers of agent washing.

James Kaplan has been consulting enterprise clients on how to adopt and extract value from cutting-edge technologies for over 25 years. His dual role gives him a unique vantage point into how CIOs and CTOs can harness AI to drive automation, increase productivity, and enhance data security.

Hosts Aaron Fulkerson and Mark Hinkle sit down with Sriram Raghavan at the AllThingsOpen.ai conference. Sriram leads a global team of over 750 research scientists and engineers at IBM Research AI. The episode explores IBM's belief that the future of AI is open source, their contribution of Docling, Data Prep Kit, and BeeAI to the Linux Foundation, and why IBM is leaning into smaller fit-for-purpose models that give developers choice over massive LLMs.

From cloud computing to infrastructure-as-a-service, Reuven Cohen has been on the cutting edge of every major technology supercycle — and he's now one of the most influential people in the agentic AI space. With the help of agents, rUv produced 10 million lines of usable code last year.