Think Like a Human, Act Like an Agent

On Wrenches and Flamethrowers: The Mechanics and Engineering Behind AI Agents

What do stress testing and overhauling of jet aircraft engines have in common with designing, building, and running AI agents? More than you might think! In a world of parallel agent workflows and stochastic testing, this conversation puts artificial intelligence and agentic workflows in the context of hands-on engineering and mechanical experiences. Join Kris, drawing on his background as a civil engineer and developer advocate, and Nena, a former airplane engine mechanic. Together in this episode, they explore how the timeless principles of mechanics - from the importance of the right tools for the job to the necessity of rigorous testing - provide a powerful framework for building better AI tooling for the real world. Kris and Nena put this theory into practice as they build and run three AI workflows live on the air: a AI news translator that uses well-known personas to make complex topics in the news more relatable and understandable, a generative AI artist that iteratively paints mountainous landscapes, and an AI writing partner that helps us write a short guide with tips on winning your favorite carnival games.

Your New Lab Partner is an AI Agent

Are the latest AI agent tools a real revolution, or are we just "trading one set of questions for another?" Kris and Eric dive into the practical realities of AI agents. They explore the idea that technology moves in "generational cycles," comparing the current AI boom to the rise of the technology stack and discussing the familiar feelings of frustration and excitement that come with any new paradigm. The discussion goes beyond specific tools to explore the deeper questions of human-computer interaction. How should we design cognitive architectures for agents? How do we navigate the "jagged edge" of AI capabilities, where models excel at one task but fail at another? They share lessons from building their own "agentic research assistant" and ponder how AI could fundamentally change the nature of research—potentially eliminating the "dreaded literature review" while raising new questions about the future of human expertise. If you're an engineer, scientist, researcher, or builder looking for a grounded, honest perspective on the challenges and opportunities of the generative AI and agentic era, this conversation is for you!