Jensen Huang’s Reindeer Games, Agent Frameworks vs. Fully Custom, and Rapid Impact vs. Technical Debt

In this week’s episode, Rob and Justin dig into the weird paralysis happening at enterprise scale. Fortune 500 companies are spending six months in high-level negotiations to build AI workflows that could be done in a week. IT departments, trained for decades to fear custom code, are watching their companies get lapped by competitors who just decided to turn the thing on. Everyone’s releasing agent frameworks, every AI company’s got one, some have more than one, and instead of clarifying things, it’s freezing people up..

There’s a massive gap between what AI can do right now and what most organizations are getting out of it. Justin calls it the capability overhang, and it’s growing. Not because the technology isn’t ready, but because of how businesses are approaching it. Rob’s got stories from the field that’ll make you feel like the guy in the manhole from Die Hard, waving his hands: “I can just do it right here. No really. Right here.”

You’ll learn what it really means to unlock AI (spoiler: it’s not about waiting for the tech to get better), why hoping for built-in solutions is a fantasy, and why Claude is now updating its own instructions when it screws up. If you’ve been wondering whether you’re behind on AI or just appropriately skeptical of the hype, this one’s for you.

Also in this episode:

Juan Garcia on Raw Data with Rob Collie

Morpheus / The Matrix

Die Hard Manhole scene