Latent Space Podcast System Prompt

Welcome to our secret podcast guest page! We’ve learned to prompt our guests to answer FAQs and give some idea of what we’re going for, excited to have you on!


  • What is Latent Space? A technical newsletter, podcast (top 10 in US Tech! in audio podcast and YouTube video formats), and community (VERY active Discord online and in person) by and for the rising class of AI Engineers. More in our /about page including quotes from Andrej Karpathy, Chris Lattner, George Hotz, Soumith Chintala, Eugene Yan, Emad Mostaque, Simon Willison, et al.

  • Can I review before you publish? Yes: All recording will be professionally edited (so you can pause to look up stuff, or rewind to restate something anytime) and will be sent to you for approval before publish (in case you want to take out something)

Guest Prompt

  • We are a podcast for software engineers building with LLMs and modern AI - both hosts are technical, though not ML experts, but we have a large/growing audience of “AI Builder” type software engineers (from OpenAI to MIT to Apple to many AI/non-AI startups), aka “AI Engineers”, who want to learn as much about the nitty gritty details of software engineering, open source, data collection, hardware constraints, model/training results and hyperparameters, any counterintuitive technical problems and dead-ends you ran into.

    • We’ll sometimes pause to introduce key concepts/projects so that people don’t get lost + get a new perspective on what they think they already know

  • Especially want to know:

    • Key discoveries/learnings that greatly impacted how you work

    • Examples and specifics over generalization

      • Anything with money, infrastructure scaling/optimization, training intuition

    • Fun anecdotes and stories about the human side of working with AI. Side tangent stories are great!

    • Underrated people who have made significant contributions that should get more attention

    • Predictions! it gets people excited + you can point back later when it comes true

    • Passionate Rants! energy and excitement makes for good audio

  • We spend less time covering: AI Safety/Legal/Regulation/Ethical debates 

    • because those arguments are often extensively covered by nontechnical podcasts and internet flamewars. We aren’t specialists in that.

    • Side tangents are ok! Tech exists in a broader context. But our listeners are here for the tech first and foremost.

  • We are aiming to be the first place many people hear about AI ideas, techniques, problems, findings, predictions, and heuristics/laws that they’ll still be talking about 6-12 months from now. Folks like Jeremy Howard, Chris Lattner, and George Hotz came on to talk about their latest work. Many AI Startups are also choosing to launch news with us, e.g. Replit, Roboflow, and more.

  • We have a very conversational tone so we do like to follow up, improvise, laugh, get loud and excited, and get to know both the person and the tech. Swearing is ok in small doses. We want people to feel like they are listening in on insider conversations in SF.

Looking forward to chat!