Latent Space
Hosted by Latent Space
A technical podcast for AI engineers, featuring deep-dive interviews with the founders and researchers building frontier models and tools.
A deeply technical interview show for practitioners in the AI space. Episodes dissect the work of top-tier guests—founders, CTOs, and chief scientists from major AI labs—grounding the conversation in recent papers, product launches, and architectural decisions. It bypasses high-level chatter to focus on the engineering challenges, strategic bets, and mental models behind building with and for AI.
“The show's primary distinction is its access to the highest-profile builders in AI, from Databricks' co-founders to OpenAI's research chief. Unlike many tech podcasts, it maintains a rigorous focus on engineering specifics—architectural diagrams, evaluation crises, and scaling laws—rather than just business or product narratives.”
Who hosts this show
Latent Space is a podcast and newsletter for AI engineers, hosted primarily by Shawn "swyx" Wang, with occasional co-hosts like Allen Park. It features in-depth conversations with leaders from top AI labs and startups like OpenAI, Databricks, and xAI, focusing on models, tools, and practical ideas. Wang, a former developer relations leader at AWS and several unicorn startups, coined the term "AI Engineer" and uses the platform to explore the intersection of software engineering and applied AI.
Credentials & credits
- Editor/Host at Latent Space
- Founder of Smol AI
- Cofounder of AI Engineer conference
- Former Head of Developer Experience at Temporal, Airbyte
- Former Developer Advocate at AWS, Netlify
- Former quantitative finance professional at Two Sigma
Other ventures
- AI Engineer (conference series)
- Smol.ai (AI company)
- DX.Tips (magazine)
- Angel investor in dev tools and AI
- The Coding Career Handbook (author)
What kind of podcast
When new episodes drop
- 01Cooking with OpenAI’s Research Chief: AGI, o1, Evals, and Scaling Laws — Mark ChenJun 25, 2026 · 41 min
- 02The Agent Cloud: Databricks’ Bet on the Future of AI — Matei Zaharia and Reynold XinJun 24, 2026 · 1h 10m
- 03AI Security After Codex and Claude Code — Zico Kolter & Matt Fredrikson, Gray SwanJun 22, 2026 · 1h 08m
- 04⚡️Every product of the future will be a living system — Ronak Malde, Trajectory.aiJun 21, 2026 · 34 min
- 05Why AI Labs With Unlimited GPUs Still Fail — Anjney Midha, AMPJun 18, 2026 · 1h 01m
- 06
- 07
- 08When AI Agents Run Businesses — Lukas Petersson and Axel Backlund of Andon LabsJun 4, 2026 · 1h 18m
Notable episodes
- 01Cooking with OpenAI’s Research Chief: AGI, o1, Evals, and Scaling Laws — Mark Chen
A unique format where the host and guest cook a meal, leading to a relaxed but deeply technical discussion on OpenAI's research strategy and the confirmation of the 'soup wars' talent story.
- 02The Agent Cloud: Databricks’ Bet on the Future of AI — Matei Zaharia and Reynold Xin
A detailed breakdown of a major new open-source agent framework from the co-founders of Databricks, illustrating the show's access to key figures during major product announcements.
- 03AI Security After Codex and Claude Code — Zico Kolter & Matt Fredrikson, Gray Swan
A timely exploration of the new vulnerabilities created by LLMs, featuring the founders of a leading AI security startup with deep roots in academic research.
- 04The Rise of the AI Engineer
The viral essay that defined the 'AI Engineer' role and serves as the philosophical foundation for the podcast and its community.
What you'll be asked on this show
Host Shawn "swyx" Wang's style is that of a peer who has done his homework. He often opens by referencing a guest's recent tweet, paper, or a public story to create an immediate, relevant entry point. He frequently asks for analogies to past computing paradigms (e.g., operating systems) to clarify complex new agent architectures and uses on-screen diagrams to ground technical discussions. To probe deeper, he often asks for contrarian opinions ('What commonly accepted ideas do you disagree with?') and the specific 'tells' they look for when hiring talent. The interviews feel less like a Q&A and more like a high-level strategy session between two well-informed practitioners.
Primarily an interview show hosted by Shawn "swyx" Wang, who is deeply knowledgeable about the AI landscape. The format often uses on-screen visuals like websites, papers, and diagrams to add context to technical discussions. While typically a standard remote interview, it sometimes employs creative formats, like a cooking show, to create a more casual atmosphere for deep conversation.
Questions Latent Space keeps coming back to
12 cataloguedIf you're going on this show as a guest, expect some version of each of these. Each note explains when Latent Space reaches for it.
origin
1- Q.01
“What's the origin story behind this company or project?”
Often asked at the beginning to establish the core mission and the problem the founders set out to solve.
controversy
2- Q.01
“Is the widely-circulated story about [X event] actually true?”
Used to confirm or debunk industry lore and get the guest's first-hand account of a pivotal or controversial moment.
- Q.02
“Are there any commonly accepted ideas in AI research that you disagree with?”
A favorite question to uncover a guest's contrarian viewpoints and strategic thinking.
process
3- Q.01
“What led you to create this specific architecture or product?”
This question seeks the technical motivation and problem statement behind a new release.
- Q.02
“When you get a new model from a lab, what's the first thing you do with it?”
A process-oriented question to understand a company's internal evaluation and red-teaming workflow.
- Q.03
“How often do you reassess your research roadmap?”
Aimed at understanding the strategic planning cycle and agility of a top-tier research organization.
craft
2- Q.01
“Does this new concept map to any ideas from past paradigms, like operating systems?”
The host uses this to create analogies that help the audience understand a novel technical concept.
- Q.02
“What's an example of something you've learned from your top users or community members?”
This surfaces specific, non-obvious insights gained from expert practitioners in the field.
future
1- Q.01
“How do you evaluate progress once models saturate existing benchmarks?”
Probes the frontier of AI measurement and the challenges of assessing superhuman capabilities.
advice
2- Q.01
“For people without a PhD, how can they develop the 'taste' to get into AI research?”
Seeks actionable advice for aspiring researchers trying to break into the field via non-traditional paths.
- Q.02
“In an interview, what are the tells that someone has the potential to be a great researcher?”
This provides insight into the hiring and talent identification process at leading AI labs.
industry
1- Q.01
“What would a lab like Anthropic or OpenAI outsource versus do in-house for a function like red teaming?”
This question clarifies the value proposition of external services for frontier model labs.
Signature segments
- · Cooking with Founders and Researchers
Topics covered repeatedly
Who gets booked here
The guest list consists of C-suite executives (CEOs, CTOs), co-founders, and lead researchers from the most influential companies in the AI industry, including OpenAI, Databricks, Anthropic, and xAI. Guests are typically there to discuss a new product, open-source release, or research direction.
- Mark Chenon Cooking with OpenAI’s Research Chief: AGI, o1, Evals, and Scaling Laws — Mark Chen
- Matei Zahariaon The Agent Cloud: Databricks’ Bet on the Future of AI — Matei Zaharia and Reynold Xin
- Reynold Xinon The Agent Cloud: Databricks’ Bet on the Future of AI — Matei Zaharia and Reynold Xin
- Zico Kolteron AI Security After Codex and Claude Code — Zico Kolter & Matt Fredrikson, Gray Swan
- Matt Fredricksonon AI Security After Codex and Claude Code — Zico Kolter & Matt Fredrikson, Gray Swan
- Anjney Midhaon Why AI Labs With Unlimited GPUs Still Fail — Anjney Midha, AMP
Where to find this show
Audience & reach
The podcast is explicitly listener-supported to remain ad-free. The host makes direct appeals for listeners to subscribe to the newsletter and podcast as the primary means of support.
Subscriber and view counts are pulled live from YouTube and re-verified on a 30-day cycle. Listener estimates for the RSS feed aren't published here unless they're host-verified.
Pitch Latent Space
s•••@•••.comFree: limited reveals · Pro: unlimited reveals + CSV export
You’re only charged when we return a verified hit.
People also ask
- What is the real name of the host 'swyx'?
- The primary host, 'swyx', is Shawn Wang. He uses 'swyx' as it represents the initials of his English and Chinese names.
- Is Latent Space still running?
- Yes, the podcast is active and releases multiple episodes per week.
- What is the format of the show?
- It is an interview-based podcast featuring conversations with founders and researchers in the AI field. Most episodes are remote video interviews, but some have unique formats like a cooking show.
- Who is the podcast for?
- It is specifically for AI engineers and technical practitioners who build AI-powered products and want to stay on the cutting edge of models, tools, and infrastructure.
- Where can I listen to the podcast?
- The podcast is available on its official website, YouTube, and major podcast platforms.
- How can I support the podcast?
- The host encourages listeners to subscribe to the free podcast and newsletter, which allows the show to remain ad-free.
Built from the show's public RSS feed, YouTube, the host's own websites, and the cited sources below. Computed and AI-extracted fields are labelled. Facts only — no private info, no fabrication, no transcripts republished.
Sources & how this page was built
This page is AI-assisted, grounded in the public sources cited below, and host-verifiable. We publish facts only; we do not republish transcripts. If anything here is wrong, the host can claim and correct the page above.Model: gemini-2.5-pro · high confidence
Podcasts like Latent Space
Thomas Brush
Thomas Brush
A podcast for aspiring indie game developers, hosted by a solo dev who interviews peers about the practical and emotional journey of making games.
The Wolf Of All Streets
Scott Melker
A daily show from trader Scott Melker breaking down crypto, macro, and trading news with market experts and industry builders.
Rusty Quill Podcasts
Rusty Quill Podcasts
A network feed from a London-based production house specializing in full-cast speculative fiction and horror audio dramas.
Satish K Videos
Satish K Videos
An interview show that deconstructs the business models of successful Indian entrepreneurs, with a sharp focus on revenue, strategy, and digital tools.
The American Business Podcast (ABP)
A live, daily business podcast that doubles as the media arm for a national, digital-first chamber of commerce.
Jay Clouse
Jay Clouse
A podcast and video series breaking down the business of being a creator, with a strong focus on building membership communities.
We In Miami Podcast
A Miami-based interview show where host Stunt Lifestyle gets hip-hop and reality TV stars to talk about their careers, feuds, and personal lives.
The Senior Health Podcast
A health show for seniors offering simple, actionable exercises and habit changes based on the philosophy of "Japan's Oldest Doctor."