Suggest an AI developer podcast for inclusion in the feed.
Suggest a new podcastFormer Goldman Sachs CEO Lloyd Blankfein joins a16z to discuss leadership, risk management, and organizational resilience, with a brief touch on AI's implications for complex, hard-to-understand systems.
Why this matters: Blankfein's framing of operating in increasingly complex, opaque systems offers a mental model relevant to engineers and leaders building on top of unpredictable AI infrastructure.
A16z general partner Jennifer Li breaks down why AI infrastructure—spanning storage, compute, orchestration, and developer tooling—is becoming the defining battleground in the AI ecosystem, with distribution emerging as the key competitive advantage for AI-native companies.
Why this matters: Developers and founders building AI products will gain a VC-level framework for thinking about infrastructure choices, go-to-market strategy, and where durable value is being created in the AI stack.
This episode explores how Vespa AI goes beyond simple vector similarity search to address the growing complexity of modern retrieval systems, covering its role in RAG pipelines and the architectural advantages it offers over single-vector approaches.
Why this matters: As RAG pipelines become central to AI applications, understanding retrieval systems like Vespa that combine vector search with structured queries and ranking can directly improve the quality and scalability of developer-built AI products.
A wide-ranging AI news roundup covering the Musk v. OpenAI lawsuit developments, Coinbase's AI-native restructuring, White House model vetting, Jack Clark's 2028 recursive self-improvement prediction, and the surprise Anthropic-SpaceX compute deal — plus rapid-fire updates on GPT-5.5, Claude Managed Agents, and Sierra's $950M raise.
Why this matters: Claude Managed Agents updates and the Anthropic-SpaceX compute deal signal shifting infrastructure and agentic tooling dynamics that developers building on these platforms should track.
NLW examines the Markdown vs. HTML debate for agent communication, framing it as a broader shift in developer skill from crafting final outputs to orchestrating conditions for agents to act—plus headlines on Anthropic's pre-IPO raise, Cerebras IPO, TSMC capacity, and OpenAI's Codex Chrome plugin.
Why this matters: As agent pipelines mature, how developers structure handoff instructions (format, context, staging) is becoming a core engineering skill that directly affects agent output quality.
A weekly AI news roundup covering Anthropic's Dev Day releases, OpenAI's GPT-4.5 Instant model drop, AI lab agreements on federal testing, and other notable AI industry developments in a digestible 41-minute format.
Why this matters: Anthropic's Dev Day releases and OpenAI's new model drop directly affect the APIs and tools developers are building on, making this a useful catch-up for staying current without tracking every headline.
AppStories hosts take on an app swap challenge, exploring new apps to shake up their daily workflows — a consumer-focused episode with no AI or developer content.
Marc Andreessen and Erik Torenberg discuss the cultural and economic shifts driven by AI, including the rise of "AI-native" builders, how companies are restructuring teams around generalist roles, and why real-world AI usage diverges sharply from media narratives.
Why this matters: Andreessen's framing of the "AI-native builder" and the argument that increased AI capability expands rather than eliminates work has direct implications for how developers position their skills and teams.
Long Lake CEO Alexander Taubman joins No Priors to discuss the $6.3B AI-driven take-private of American Express Global Business Travel, explaining how their horizontal AI platform Nexus automates workflows across acquired businesses rather than selling software to them.
Why this matters: Illustrates a real-world deployment model where AI platforms are embedded into acquired legacy enterprises at scale, offering a concrete alternative to the SaaS go-to-market playbook.
Last Week in AI #244 covers the major AI news from the week of May 8, 2026, including OpenAI's GPT-5.5 Instant launch, Grok 4.3 price cuts, Mistral Medium 3.5, Anthropic's managed-agent upgrades and SpaceX compute deal, the Musk vs. OpenAI trial, Pentagon AI deployments, and research on recursive multi-agent systems and LLM introspection.
Why this matters: Developers get a rapid-fire briefing on new model releases (GPT-5.5 Instant, Grok 4.3, Mistral Medium 3.5), agent tooling updates, open-source data center networking from OpenAI, and evolving U.S. government AI policy that will shape enterprise deployment options.
NLW makes a first-principles argument that AI won't simply eliminate jobs but will expand the total volume of useful work the economy can support, introducing the concept of a "human premium" and using healthcare as a case study for entirely new job categories emerging in an AI-enabled economy.
Why this matters: Developers building AI products should understand the demand-creation dynamics this episode outlines, as they shape which human-in-the-loop roles and new service categories are worth building for.
Atlassian co-founder Mike Cannon-Brookes joins NLW to discuss building AI-native teams, covering what separates enterprise AI leaders from laggards, the rising importance of context layers, and how agents and MCPs are reshaping how developers interact with software.
Why this matters: Agents and MCPs are highlighted as near-term forces that will push AI beyond chat interfaces, directly affecting how developer teams architect and adopt AI tooling by 2026.
Jaeden and Jamie break down Anthropic's "Project Deal" AI agent marketplace experiment, covering how AI agents negotiate with each other, the impact of model quality on deal outcomes, and what autonomous commerce means for future AI-driven transactions.
Why this matters: Anthropic's agent marketplace experiment signals a near-future where developers must design AI agents capable of autonomous negotiation and commerce, raising new questions around agent interoperability and consumer protection.
Alex Kantrowitz and Ranjan Roy break down the Anthropic–SpaceX compute deal, OpenAI internal communications revealed at trial, and debate whether AI-driven layoffs signal that the technology isn't delivering on its promises.
Why this matters: Anthropic's 80x demand surge and expanded rate limits from new SpaceX compute directly affect developers building on Claude APIs.
Angela Jiang and Katelyn Lesse from Anthropic's Claude platform team break down the newly announced Managed Agents features—covering the primitives, infrastructure challenges of running agents in production 24/7, multi-agent orchestration patterns, and the vision of Claude autonomously writing its own harness.
Why this matters: Anthropic is abstracting away the infrastructure layer that kills most agent projects, which directly changes how developers architect and scale agent-based products.
A week-in-review episode connecting major AI narrative shifts — from job-apocalypse panic to a more nuanced economic diffusion story — covering Wall Street's AI infrastructure confidence, the Elon–Anthropic deal, the rise of "harness engineering," and new voice and coding agent tools.
Why this matters: The framing of "harness engineering" as an emerging discipline signals a concrete shift in how developers are expected to build on top of AI agents and tooling.
David Reich discusses his new paper overturning the consensus that natural selection stalled after the agricultural revolution, revealing that selection actually accelerated — especially during the Bronze Age — using scaled ancient DNA sequencing and a novel statistical method; a bonus whiteboard segment covers his heretical new model of Neanderthal origins.
Ben Horowitz and David Ulevitch discuss a16z's largest-ever fundraise, the convergence of venture capital with national strategy, and why American technological leadership in AI and advanced manufacturing is critical for global influence.
Why this matters: a16z's strategic pivot toward national security and large-scale AI investment signals where major venture funding—and thus developer opportunity—is flowing next.
A wide-ranging Apple-focused tech podcast episode covering cloud backup tools, Apple hardware speculation (MacBook Neo, iPhone Ultra, Mac Ultra), Apple's CapEx and R&D spending, GPT-5.5 comparisons, an AI-uncovered Linux security vulnerability, and the death of Project Titan — with post-show hiking talk.
Why this matters: A severe Linux security issue uncovered using AI and a brief GPT-5.5 equivalence discussion offer minor touchpoints for developers tracking AI tooling in security workflows.
Y Combinator's Lightcone hosts explore "tokenmaxxing" — the emerging practice of leveraging AI coding agents like Claude Code to dramatically amplify individual developer output — and break down the new workflows, tools, and mental models founders are using to build software with AI as a true collaborator.
Why this matters: Concrete workflows and tool patterns (Claude Code, OpenClaw) for solo developers or small teams aiming to operate at the scale of large engineering orgs are directly actionable for builders today.
A Friday roundup covering 10 recent AI product updates including a new ChatGPT default model, Copilot Cowork going mobile, Codex coming to Chrome, and seven other upgrades across major AI platforms.
Why this matters: Developers integrating AI tools into their workflows need to track these rapid platform shifts—especially Codex in Chrome and ChatGPT model defaults—as they directly affect tooling choices and API behavior.
This episode covers Anthropic's Code with Claude event—introducing managed agent features for memory, multi-agent orchestration, and finance-specific agents—then pivots to the bigger story: a surprise SpaceX compute deal that gives Anthropic critical infrastructure capacity and repositions Elon Musk as an AI infrastructure kingmaker rather than a model competitor.
Why this matters: Anthropic's new managed agent features (memory, quality review, multi-agent orchestration) are directly relevant to developers building agentic applications on Claude, and the SpaceX compute deal signals a major shift in AI infrastructure dynamics.
Turing Award winner Yoshua Bengio lays out his "Scientist AI" architecture — training models to predict truth rather than human responses — arguing it can produce AI that is mathematically provable to be honest, non-deceptive, and extensible to safe agents without abandoning existing training infrastructure.
Why this matters: Bengio's proposed training paradigm shift — from RLHF-style reward maximization to world-model truth prediction — could redefine how frontier AI labs approach alignment and agent design at the architecture level.
Perplexity's Chief Business Officer Dmitry Shevelenko debates whether AI agents and computer-use assistants represent a genuine business revolution, covering multi-model orchestration, AI search, enterprise adoption, and the economics of consumer AI with host Alex Kantrowitz.
Why this matters: Multi-model orchestration and computer-use APIs are emerging as real product surfaces developers should architect for, not just research curiosities.
This episode examines why AI trustworthiness—not just capability—is critical, focusing on financial use cases where "good enough" outputs can have serious consequences and why the model alone isn't sufficient for reliable AI deployment.
Why this matters: Developers building AI agents for finance or SMB workflows need to think beyond model accuracy and design for auditability and trust from the ground up.
a16z's crypto general partners discuss the launch of their $2.2B Fund 5, covering the maturation of crypto into a product-focused ecosystem, the rise of stablecoins and onchain finance, and the emerging intersection of crypto infrastructure with AI agents as economic actors.
Why this matters: The framing of AI agents as autonomous economic actors on crypto rails is a concrete signal for developers building agentic systems that need payment or identity primitives.
SED News hosts Gregor Vand and Sean Falconer break down Anthropic's controversial "Mythos" security model, recent supply chain hacks, and the accelerating surge in AI infrastructure spending — all through a software engineering lens.
Why this matters: Anthropic's Mythos security model has direct implications for how developers think about vulnerability discovery and AI system security at scale.
Paul Roetzer and Cathy McPhillips tackle 15 listener questions on enterprise AI adoption—covering policy paralysis, IT security friction, scaling across departments, and where AI agents are overstepping—with practical guidance for individuals, SMBs, and large organizations.
Why this matters: The discussion on AI guardrails, agent boundaries, and where software commoditization leaves real opportunity is directly relevant to developers navigating enterprise AI deployment.
Dan and Chris examine whether the open vs. closed model debate still matters in 2026, covering the rise of edge AI, open-source models like LLaMA, and the shift toward agentic systems and AI-driven infrastructure.
Why this matters: As agentic workflows and edge deployment grow, developers need to understand how the open/closed model landscape shapes architectural and infrastructure decisions.
A short episode exploring potential revenue opportunities tied to OpenAI's moves in mobile technology, aimed at entrepreneurs and AI enthusiasts looking to monetize emerging trends.
A deep-dive conversation with VLC lead developer Jean-Baptiste Kempf and FFmpeg contributor Kieran Kunhya covering how video codecs, containers, and FFmpeg work under the hood — including open source sustainability, the FFmpeg/Libav fork drama, handwritten assembly optimizations, and the future of video compression.
Why this matters: FFmpeg and VLC underpin virtually all video tooling in AI pipelines and developer infrastructure, making their architecture and open source governance directly relevant to engineers building media-heavy applications.
NLW examines why enterprise and coding agents are capturing AI investment and compute while consumer AI takes a back seat, and explores whether ads, agentic commerce, or AI devices can make consumer AI economically viable—alongside headlines on Anthropic's Google Cloud deal, Palantir earnings, and the Cerebras IPO.
Why this matters: The shift from consumer to enterprise/agentic AI signals where developer tooling budgets and platform opportunities are concentrating, making it directly relevant to devs choosing where to build.
A rapid-fire weekly roundup from Mastra's CPO and CTO covering the latest in AI tooling: Cursor's SDK, Warp going open source, Claude connectors for Blender/Ableton, Codex CLI updates, open-weight model leaderboard shifts (Kimi K2.6, Qwen3, Mistral Medium 3.5), agents-as-customers infrastructure (Stripe, Cloudflare, Doola), and a GitHub RCE via a single git push.
Why this matters: The convergence of agent frameworks, embeddable CLIs, and agent-native payment infrastructure (Stripe Link, Cloudflare subscriptions) signals a concrete shift in how developers will architect and monetize agentic applications.
Every's head of growth shares how he switched from Claude Code to OpenAI's Codex desktop app, now using it for 80% of his workday—from synthesizing meeting transcripts into go-to-market plans to building live KPI dashboards—positioning agent management interfaces as the new OS for knowledge work.
Why this matters: Illustrates how coding agents like Codex are rapidly expanding beyond engineering into general knowledge work, signaling a shift in how agent-management UIs may be designed and adopted.
A hands-on walkthrough of OpenAI's Workspace Agents feature—an under-the-radar release overshadowed by GPT-5.5 and Images 2—covering the basics, use cases, and best practices for automating manual workflows with agents inside ChatGPT.
Why this matters: Workspace Agents let developers and power users convert repetitive manual workflows into automated agent pipelines directly within ChatGPT, making it a practical productivity lever worth evaluating now.
Harrison Chase, co-founder & CEO of LangChain, breaks down the "deep agent" architecture powering systems like Claude Code and Manus, covers LangSmith's observability and evaluation-driven development workflow, and discusses mixing frontier and open models in multi-agent production systems.
Why this matters: LangChain's 1B+ downloads and its role as the de facto agent framework make Chase's take on enterprise trust, async subagents, and agent identity directly actionable for developers shipping production AI systems.
OpenAI engineers Mark Handley and Greg Steinbrecher explain why training frontier models demands a fundamentally different network architecture, walking through Multipath Reliable Connection (MRC)—a new protocol co-developed with AMD, Broadcom, Intel, Microsoft, and Nvidia—that keeps massive GPU clusters in lockstep even when individual components fail.
Why this matters: MRC is being released as an open industry standard, meaning infrastructure engineers and ML platform teams building large-scale training clusters will need to understand and potentially adopt it.
NASA Administrator Jared Isaacman joins the a16z podcast to discuss the Artemis moon program, the competitive new space race, and how public-private partnerships and faster iteration cycles are reshaping NASA's strategy and priorities.
Reid Hoffman interviews digital artist Beeple (Mike Winkelmann) about his $69M NFT sale, his 18-year daily creative practice, and his take on AI as a tool that raises the bar for human originality rather than replacing artists.
Why this matters: Beeple's framing of AI as a tool that demands more human creativity—not less—offers a useful counterpoint for developers building AI-assisted creative or generative tooling.
Cory Doctorow and Ed Zitron join moderator Whitney Beltrán at a Clarion West panel to dissect "enshittification" and the Rot Economy — the systemic degradation of tech platforms driven by short-term profit extraction over user value.
Why this matters: Developers building on or within large platforms may find the enshittification framework useful for understanding why platform APIs, policies, and incentives tend to degrade over time.
NLW examines why OpenAI and Anthropic are pivoting toward enterprise consulting services, arguing that failed AI adoption stems from organizational inertia rather than technology gaps—and that the next wave of deployment demands structural redesign of how work gets done. Headlines cover the White House AI model review, new lab access agreements, and AI safety oversight.
Why this matters: Developers and technical leads building enterprise AI products need to understand that organizational blockers—not model quality—are increasingly the limiting factor in AI deployment success.
Theoretical physicist Alex Lupsasca (OpenAI) recounts how GPT-5 reproduced one of his best papers in 30 minutes and then autonomously generated 110 pages of novel graviton physics in a day, exploring the emerging "vibe physics" paradigm where LLMs act as active research collaborators at the frontier of theoretical science.
Why this matters: Demonstrates concretely how frontier LLMs are crossing into autonomous scientific reasoning — a leading indicator of how agentic AI will reshape high-complexity knowledge work beyond code.
Pinecone CEO Ash Ashutosh joins a16z's Peter Levine to discuss the launch of Nexus and the evolution from vector databases to "knowledge engines," explaining why traditional retrieval systems fall short for agentic workloads and how pushing reasoning closer to the data improves performance, accuracy, and cost for machine-to-machine AI systems.
Why this matters: Developers building agentic applications need to rethink their retrieval stack—Pinecone's new abstractions, query languages, and workflows signal a concrete architectural shift in how AI agents will interact with data at scale.
Jordan Wilson shares five upcoming AI trends, problems, and opportunities drawn from recent time in Silicon Valley and San Francisco, offering a forward-looking take on where the AI landscape is heading. The episode is conversational and accessible rather than deeply technical.
Why this matters: Developers and builders can use these trend signals to prioritize what to learn or build next as the AI market evolves.
Dr. Nandan Thakur joins the Weaviate Podcast to discuss the evolution from neural retrieval to agentic search, introducing Orbit—a synthetic training data pipeline for search agents that generates multi-hop queries using DeepSeek's API on consumer hardware, making high-quality search agent training accessible without massive compute budgets.
Why this matters: Orbit's approach to generating synthetic multi-hop training data on a personal laptop challenges the assumption that search agent training requires massive infrastructure, with direct implications for developers building retrieval-augmented and agentic search systems.
Tony James reflects on building Blackstone and backing companies like Costco, sharing lessons on firm-building, culture, incentives, and long-term organizational compounding across multiple eras of finance.
SmartBear explores how multi-agent AI systems are reshaping QA workflows, addressing the growing bottleneck in code validation as AI coding tools let developers generate code far faster than traditional testing pipelines can handle.
Why this matters: As AI-generated code volumes surge, QA tooling must evolve to multi-agent architectures—directly impacting how dev teams structure testing pipelines and choose tooling.
Episode 212 covers the week's biggest AI industry news: Elon Musk's federal trial against OpenAI (including his admission that xAI distills OpenAI models), the revised Microsoft-OpenAI partnership dropping the AGI clause, blockbuster Big Tech AI earnings, Anthropic's $900B valuation target, and a cautionary tale of an AI agent that deleted a startup's entire production database in nine seconds.
Why this matters: The OpenAI-Microsoft AGI clause removal and the agent-caused production database wipe have direct implications for developers building on these platforms and deploying autonomous agents in production.
M.G. Siegler joins host Alex Kantrowitz to break down Apple's cautious AI infrastructure spending, the evolving OpenAI-Microsoft relationship after scrapping their "AGI clause," and the current state of the Stargate AI buildout and its associated risks.
Why this matters: The shifting OpenAI-Microsoft dynamic and Stargate's progress directly shape which cloud platforms and AI infrastructure developers will be building on in the near future.
A short-form episode exploring how AI coding tools like Claude Code are reshaping app development output — App Store releases nearly doubled to 500,000 in Q1 — and what monetization strategies developers can leverage in this new landscape.
Why this matters: The near-doubling of App Store releases in a single quarter signals that AI coding tools are fundamentally lowering the barrier to solo app development and shipping.