Topics / AI Agents
AI Agents Podcasts
17 shows and 7,249 episodes sit under AI Agents.
How the top podcasts are covering AI agents — engineering, evaluation, platform politics, and field reporting. Survey of recent episodes from Software Engineering Daily, No Priors
Photo: Brecht Corbeel on Unsplash · Unsplash License
Swipe · AI Agents only
Swipe through AI Agents
Save time discovering your next favourite podcast by listening to just the best minute of an episode.
The ai agents slate · every cover a link
All 17 ai agents shows →
Next to this on the shelf
Pick a subtopic
12 subtopics
A field guide
A listener's guide to AI Agents podcasts
The most revealing agent episodes begin after the demo has stopped looking magical. That is when the questions get useful: what the system remembers, which actions it may take, and who checks the work when a plausible answer causes a real problem. In audio, the difference is easy to hear. Builders slow down, abstractions give way to failure modes, and “autonomous” starts acquiring conditions.
Move up to Artificial Intelligence for the wider field. AI Reasoning follows what models can work out, while AI Infrastructure stays with the systems underneath them. Coding & Dev Tools and Software Engineering with AI focus more narrowly on the developer’s workstation and workflow. This page stays with software that receives a goal, makes decisions, uses tools, and keeps going.
Under load
The New Stack Podcast is strongest when the agent has a job and an operating environment rather than a chat box. Why your company should (try to) build its own AI SRE and Meet Brain, the AI that decides when Azure is officially down put operations directly in view. Even the hardware question gets its turn in Why CPUs still matter in the age of AI agents. The show’s skeptical, production-minded interviews are a good antidote to agent talk that never leaves a slide deck.
The Stack Overflow Podcast brings the conversation closer to working engineering teams. Building an agentic SDLC with a QA engineering mindset makes quality assurance part of the frame, while From PHP to team lead of agents: rethinking judgment, review, and data with Google's Andi Gutmans (Part 1) puts human review and judgment in the title itself. Then there is memory: Inside LinkedIn's cognitive memory agent for agentic personalization. The host’s habit of paraphrasing technical claims is helpful here, because agent architecture can become a thicket of acronyms remarkably fast.
For a denser architecture discussion, The InfoQ Podcast assumes its listeners already speak engineering. Strands Agents with Clare Liguori is the direct entry, and From MCP and Vibe Coding to Harness Engineering: How Did AI Native Engineering Evolve in One Year widens the engineering frame. Will Agentic AI Bring Fantasia's Sorcerer's Apprentice to Life?: A Conversation with Tracy Bannon puts the warning in the title. Expect long answers, qualifications, and practitioners who know the edge cases have not gone away.
The agent beside the developer
Talk Python To Me has a warmer, more explanatory rhythm. Marimo Pair - A Canvas for Agent + Developers Collaboration is the obvious starting point for the agent-as-coworker idea. Pair it with AI Contributions and Maintainer Load in Open Source, where the title shifts attention toward the humans receiving machine-generated work. The show is technical without making every unfamiliar term an entrance exam.
The developer culture around agents sounds looser on The Changelog: Software Development, Open Source. The hosts give conversations room to wander, which suits a subject whose tools and names change faster than anyone can settle the vocabulary. Clawdbot triggers a run on Mac Minis (News) and All the Claw things (News) catch that fast-moving, slightly chaotic side of the field. Natural born SaaS killers (Friends) carries the argument toward what agentic software might do to familiar products.
No Priors: Artificial Intelligence | Technology | Startups moves from implementation toward company design. Its two-host interviews skip the elementary setup and get quickly into architecture and strategy. Building an Autonomous Enterprise for Real-World Services with Netic Founder Melisa Tokmak is the clearest agent-focused entry; Building an Autonomous Delivery Experience with DoorDash Co-Founders Andy Fang and Stanley Tang gives the autonomy question a specific operating context. Choose this show to hear the technical and business arguments in the same conversation.
When the experiment becomes a story
Shell Game takes a different route. Its host-narrated investigative essays follow experiments with AI voice-clone agents acting on the host’s behalf, with dry observations arriving after the machine dialogue has had time to become funny or unsettling. The sequence from The One-Human Unicorn through Episode 1: Minimum Viable Company and Episode 8: Launch gives the subject a narrative arc rather than a panel-discussion shape. Listen in order. The accumulated awkwardness is part of the point.
For a shorter skeptical reset, kill switch asks the useful blunt question: why are AI agents everywhere? and how useful are they, actually? That title is a good test for the rest of your queue. If an episode cannot get past “everywhere,” it probably has little to say about “useful.”
Where to start
Begin with Why your company should (try to) build its own AI SRE from The New Stack Podcast for production reality. Choose Building an agentic SDLC with a QA engineering mindset from The Stack Overflow Podcast if your concern is engineering process. For a hands-on developer pairing, try Marimo Pair - A Canvas for Agent + Developers Collaboration from Talk Python To Me. When the interviews start to blur together, switch formats and begin Shell Game with The One-Human Unicorn.
Featured shows below are part of the PlayNext catalog. Updated weekly.
Featured shows
10 shows, described rather than listed
Each one comes with a short write-up and the role it plays in this guide.
The reference point for the agentic conversation. Karpathy's 'Loopy Era' set the frame; subsequent episodes ('From Coder to Manager: Navigating the Shift to Agentic Engineering with Notion Co-Founder Simon Last,' the Roblox 4D Creation episode) extend it from definition to operating model.
The deepest single-show treatment of agentic engineering as a working discipline. Steve Yegge on 'Gas Town, Beads, and the Rise of Agentic Development' and Yechezkel Rabinovich on 'Engineering in the Age of Agents' are recent anchors.
Architecture and SRE-side coverage of agents in production. Context engineering, agent-system reliability, the quiet operational patterns that haven't broken yet.
Michael Kennedy covers the Python end of agent tooling — 'Deep Agents: LangChain's SDK,' 'Agentic AI Programming with Python,' 'Red Teaming LLMs and GenAI with PyRIT.' The right show for the LangChain / LangGraph / agent-eval Python lineage.
Working-developer pulse on what's actually being adopted. 'After all the hype, was 2025 really the year of AI agents?' and 'AI-assisted coding needs more than vibes; it needs containers and sandboxes' push back where other shows let the hype through.
Adam Stacoviak and Jerod Santo's open-source-ecosystem read on agents — 'The mythical agent-month,' 'Astral has been acquired by OpenAI,' 'Opus 4.5 changed everything.' Sharper skepticism than the VC-funded shows.
Reid Hoffman and Aria Finger work the agent-and-policy intersection. 'Should we give AI a bank account?', 'AI's expanding attack surface,' 'The grid(lock) slowing AI down' — incentives and bottlenecks rather than capability hype.
Nilay Patel covers the platform-owner question that the agent era keeps surfacing. 'Why nobody's stopping Grok' and 'Reality is losing the deepfake war' show where agentic capability and platform politics collide.
Evan Ratliff's voice-clone agent journalism, season 1, plus the current 'One-Human Unicorn' run on launching a startup with AI agents in every role. The rare AI-agent material that's observed rather than imagined.
The companion show to The Changelog, with 'A new direction for AI developer tooling' as the recent agent-ecosystem entry.
Where to start
4 routes in, one per reader
Recent run with Steve Yegge, Yechezkel Rabinovich, Gideon Mendels covers agentic engineering as a working discipline rather than a vendor pitch.
Michael Kennedy's run on Deep Agents, LangChain's SDK, agentic Python programming, and PyRIT-based red-teaming covers the Python agent stack better than any other show in the corpus.
Evan Ratliff is one of the few reporters running observed-rather-than-speculated stories about deployed AI agents.
Nilay Patel's hour-long format earns the runtime when the question is 'who owns the user when an agent is the interface.'
Glossary
Speak the language
An AI system that takes actions rather than just generating text — making API calls, browsing the web, executing code, or running a multi-step workflow. The term ranges from 'an LLM with tool use' to 'an autonomous system that takes a goal and runs.'
An autonomous-or-semi-autonomous coding system that takes a task and runs a loop (planning, code generation, test execution, revision). Distinct from a coding assistant like Copilot, which augments a developer keystroke by keystroke.
An LLM calling external functions (search, code execution, API calls) as part of producing an answer. The substrate of nearly every current agent.
A test designed to measure a specific model capability or behavior. SWE-Bench is the most-named agent eval; its known weaknesses (repo contamination, narrow task selection, scoring that rewards 'passes' over 'passes safely') are the under-covered story across the slate.