What topics and guests does the Latent.Space podcast cover?

Updated

Latent.Space is the AI Engineer newsletter and technical AI podcast that focuses on how leading labs build Agents, Models, Infra, and AI for Science. If you want to know whether its content depth matches your interests, the short answer is: it is aimed at people who want to follow AI engineering at a technical level, and its guest list includes well-known figures such as Greg Brockman, Andrej Karpathy, George Hotz, Simon Willison, and Soumith Chintala.

Topics covered

According to the publication's own description, the show and newsletter center on how leading labs build:

  • Agents — how agent systems are designed and built
  • Models — model development and related engineering work
  • Infra — the infrastructure behind AI systems
  • AI for Science — applying AI in scientific contexts

This framing matters if you are deciding whether to follow it: the content is positioned around AI engineering practice and the people building these systems, not general AI news or beginner tutorials.

Guests mentioned

The site's description highlights appearances from:

Guest Context
Greg Brockman Named in the publication's highlights
Andrej Karpathy Named in the publication's highlights
George Hotz Named in the publication's highlights
Simon Willison Named in the publication's highlights
Soumith Chintala Named in the publication's highlights

The description points to the about page for highlights from these guests, so if a specific name is the reason you are interested, that page is the place to confirm what was actually discussed.

Depth and tone

Two reader testimonials quoted on the site give a sense of what to expect:

  • One reader calls it "Best AI and LLM commentary."
  • Another says the hosts "will not only keep you up-to-date about AI Engineering but also entertained — they're as funny as they are informed."

So the intended balance is technical depth plus entertainment, rather than dry documentation or pure news recaps.

Who it fits

This content is a reasonable match if you:

  • Want to follow frontier AI engineering developments
  • Are interested in how labs actually build agents, models, and infrastructure
  • Prefer long-form technical conversation over short news items

It is less suited if you are looking for introductory explanations of AI concepts or non-technical coverage.

Practical notes

The site states it has over 203,000 subscribers and offers a Subscribe option. The page also notes that the site requires JavaScript to run correctly, so if content does not load, enabling JavaScript or unblocking scripts is the first thing to check.