Latent Space

We picked Latent Space because it is the sharpest technical conversation in AI. The hosts ask working engineers the questions practitioners actually wrestle with—when to abandon RAG for grep, how agent scaffolding simplifies as models improve, how to balance latency against cost in a live clinical setting. The show trades hype for specifics: you hear about token economics, modality trade-offs, and the mechanics of deploying large models at scale. If you build with AI, or want to understand the people who do, this feed earns its place in your rotation.