These seven podcasts explore those questions through news, research, and the work of teams. Their hosts bring different backgrounds and concerns to those discussions.
The AI Daily Brief
Nathaniel Whittemore, creator and host of The AI Daily Brief, follows the announcements and debates shaping AI. The daily format, around 30 minutes long, lets him concentrate an episode on a subject and its implications.
New models are examined alongside questions of price, reliability, and use. Agents lead into individual work, collaboration, and deployment in the real world. Discussions about business and policy show how economic and regulatory constraints also shape AI labs' decisions. Whittemore hosts most episodes alone and develops his analysis in segments such as The Take. His daily analysis follows a subject as announcements and responses develop.
Practical AI
Daniel Whitenack and Chris Benson talk with engineers, founders, and technical leaders about implementing AI. The question is often how to move from a demonstration to a system that can be deployed and maintained.
An interview about enterprise agents considers the role of an AGENTS.md file alongside security, permissions, scale, and compliance. Conversations about computer-use agents and system architecture extend that work: which components to bring together, how to make them work, and where to put checks? Episodes take around 50 minutes. Security and maintenance are part of those engineering decisions once a system needs to serve a company.
The TWIML AI Podcast
Sam Charrington interviews researchers and technical leaders about AI methods, data, and evaluation. Conversations, around an hour long, devote substantial attention to explaining the research problem.
A discussion about voice agents examines the delays, interruptions, and inference costs that affect an interaction. With Justin Johnson, world models raise questions about representing an environment and evaluating the result. With Chris Potts, growing token use leads to examining what additional spending actually delivers.
Subjects also extend into vision, image generation, and connections between physics and new architectures. TWIML helps explain what researchers are trying to improve, the approaches they test, and the limits that remain.
Latent Space
Latent Space focuses on AI engineering: the work of people building products and tools with models. Swyx leads the podcast, with Alessio Fanelli, Vibhu, and other co-hosts depending on the format.
Long technical interviews take around 1 hour 45 minutes. That space allows a model to be connected to the infrastructure serving it, developer tools, and product decisions. Conversations with OpenRouter and Modal explore model access and compute infrastructure; others follow agents, video, or systems built for science. The appeal is in hearing the people behind those projects explain their choices. Models, inference, tools, and product emerge as problems to solve together.
No Priors
Inference speed can determine which products become possible; a company's data, what its agents can actually use. No Priors follows those connections between technical choices and market hypotheses. Investors Sarah Guo and Elad Gil talk with researchers, founders, and executives in interviews of about 40 minutes. With Stefano Ermon, co-founder of Inception, they examine model architectures through cost and speed. With Eon's founders, access and security accompany the use of data by agents. Michael Lee discusses using AI to transform established businesses. The program looks at the conditions under which a technology can support a business: available infrastructure, the value of data, organization, and adoption. The investors' perspective is part of its identity.
The Cognitive Revolution
Nathan Labenz follows AI capabilities and their technical, economic, and political consequences. Long interviews run around 1 hour 50 minutes; the feed also includes collective analysis formats such as AI:AM.
The work of agents can lead to concrete questions about memory and information retrieval, as in the conversation with MongoDB's Pete Johnson. With Wade Foster, CEO of Zapier, workflows raise the question of which steps require reasoning and which can remain deterministic.
Discussions about safety and governance extend that thinking to institutions able to evaluate and control systems. The interview with Anton Leicht explores the balance of power between labs and states. These positions are discussed with the guests; the podcast allows listeners to follow the arguments connecting a technical capability to a societal choice.
The Artificial Intelligence Show
Paul Roetzer and Mike Kaput translate AI news into questions for companies and their teams. Roetzer is founder and CEO of SmarterX and the Marketing AI Institute; Kaput is Chief Content Officer.
The main weekly news debrief takes around an hour and a half. New capabilities are connected to work, skills, and deployment decisions. Q&A editions examine budgets, sensitive data, human approvals, and vendor contracts.
Interviews about company transformations give those questions another form. The Baptist Health example includes training and peer coaching in a regulated sector. Adoption becomes more concrete through the people learning, those making decisions, and the changes in a team's work.
Find an explanation when you need to use it
A discussion about agent memory or enterprise adoption can be useful long after publication. Kontenu helps you find it by subject among your saved YouTube videos and the episodes in your public Spotify playlists. You can read available summaries and export your own Markdown notes as a record of what matters to you.
Try the app with your own content, then return to an explanation when your work brings you back to it. The beta is free, by invitation.