In this episode, we’re diving into how eng leaders can navigate ambiguity & bring ambitious AI research into production with Nick Gillian, Founder & CTO @ Archetype AI. Nick shares lessons from building sensor-based technologies at Google, including Project Soli, and how those experiences shaped his approach to building physical AI. He breaks down two frameworks—the two-bucket approach & the steel cable—for balancing exploration with predictable execution while delivering real customer value. Additionally, Jerry and Nick discuss how to pair researchers with software engineers, create the architecture for fast iteration, and avoid the handoffs that slow down zero-to-one development. Lastly, they cover Archetype’s early bet on open-source language models & why focusing on your core expertise can be a powerful technical and business decision.
The CAFE(S) framework is a research-backed definition of context quality for AI agents.
DX collaborated with researchers at Capital One, Google, and UVic to give teams a shared standard for reviewing agent context by breaking it into five properties: clarity, actionability, fidelity, efficiency, and security.
Your agent is only as good as its context and the CAFE(S) framework establishes a shared language around context to help you better understand how & why agents go wrong.
Podden och tillhörande omslagsbild på den här sidan tillhör
The Engineering Leadership Community (ELC). Innehållet i podden är skapat av The Engineering Leadership Community (ELC) och inte av,
eller tillsammans med, Poddtoppen.