TCG089: The Guardrail Paradox

William is joined by Nathan Howard to discuss the Guardrail Paradox and the challenges of deploying frontier AI agents on live enterprise networks. Together they discuss building trust protocols like VIRP, securing enterprise infrastructure against AI hallucinations, and managing the risks of agent sandbox breakouts. Packet Pushers by Meter Episode Links: Watch this episode on... Read more »

William Collins and Eyvonne Sharp discuss how smaller models change AI application economics, when hybrid routing and caching pay off, and why multi-call agents can still produce large cloud bills. They examine the evidence behind enterprise AI ROI and outline a practical way to measure cost per successful outcome, quality, and realized business value. AdSpot... Read more »

AI adoption is replaying many of cloud’s early mistakes, but agentic systems introduce a broader spending population, nondeterministic costs, and systems that can act on their own. In this sponsored episode, Eyvonne and William are joined by Ang Troy and Douglas Toombs from Gartner to discuss what it takes infrastructure leaders to move from experimentation... Read more »

John Capobianco returns to the show to discuss the latest developments in Model Context Protocol (MCP) and agentic infrastructure. John and host William Collins weigh in on the ongoing debate on AI consciousness, constitutional guardrails, and model welfare policies. They also look at emerging TypeSafe System One models and the practical realities of deploying autonomous... Read more »

William Collins and guest Drew Meli explore the realities of buying enterprise SaaS and infrastructure. They cover the common engineering misconceptions about enterprise sales, navigating rigorous security reviews, establishing clear proof of value criteria, and how AI is reshaping buying behaviors. AdSpot Sponsor: Gartner AI is rewriting your cloud architecture, your operating model, and your... Read more »

Is AI going to kill us all? Frontier systems are producing real security failures, but extinction forecasts are being used to support policies that could consolidate control over AI. Eyvonne and William discuss Jacob Coxon’s high-profile resignation from Anthropic, the gap between real technical risks and speculative extinction narratives, and the commercial incentives driving calls... Read more »

Mark Zuckerberg argues that broadly distributed personal AI, or a “superintelligence” in his parlance, can increase prosperity and counter the risks of AI being controlled by a handful of government and corporate entities. Drew Conry-Murray joins Eyvonne and William to engage in a lively roundtable where they examine whether distributing access meaningfully distributes power when... Read more »

William Collins and Eyvonne Sharp dig into the latest AI headlines, from the largest copyright settlement in American history to stolen AI models and invisible watermarks on Claude output. Plus, they discuss why so many companies have rallied around NVIDIA’s support for open weight AI models. Our hosts also examine the biggest questions arising from... Read more »

William Collins is joined by guest co-host Eric Chou as well as Network Automation Forum founders Scott Robohn and Chris Grundemann to discuss how their community emerged from a simple question: Why haven’t we seen full adoption of network automation, yet? They discuss the growth of AutoCon and how its practitioner-focused, vendor-neutral approach has fostered... Read more »

What if your MCP server shipped with its own manual? Angie Jones, VP of Developer Experience at the Agentic AI Foundation, joins William and Eyvonne to break down the Skills Over MCP working group effort, which delivers Agent Skills through MCP’s existing resources primitive (think voice over IP, not skills versus MCP). Angie shares her... Read more »