Joanna Stern joins Nilay Patel to talk through a year of living with AI, the gap between flashy demos and useful products, and the privacy costs hidden inside convenience. The conversation also turns to her leap from The Wall Street Journal to her new venture, New Things, and what it takes to build an independent tech-media business around YouTube, newsletters and a mainstream audience.

Eric Ries argues that good companies rarely collapse because of competition; they are more often hollowed out by governance, incentives and financial pressure that turn success into a liability. Drawing on examples from Anthropic, Novo Nordisk, Cloudflare and Philip Morris’s disastrous purchase of Vectura, he makes the case that founders must build mission protection into both culture and corporate structure before it is too late.

A sprawling Friday news roundtable connects Anthropic’s massive compute deal with SpaceX, the OpenAI trial’s newly public text messages, and a wave of AI-era layoffs. Beneath the gossip and spectacle runs a sharper argument over whether the boom is driven by real product demand, IPO maneuvering, and a genuine rewiring of how companies operate.

Anthropic’s platform leads sketch how AI infrastructure is shifting from a bare completion endpoint to managed, long-running agents with memory, tools and cloud scaffolding. They argue that the real bottleneck is no longer prompt craft but the dreary work of production infrastructure, and imagine a near future where Claude chooses its own model, spins up sub-agents and compresses work down to an outcome and a budget.

Perplexity’s chief business officer argues that the industry’s rush toward agentic “super apps” is less a pivot than an attempt to monetize AI where it already matters most: work. The conversation weighs whether computer-controlling assistants are durable businesses or just the next novelty spike, and what it means to trust them with calendars, email, taxes, and the modern payroll budget.

A conversation about AI in schools moves past cheating panics to the deeper question of what education is for when machines can generate polished work on demand. Teachers and researchers describe a fractured landscape in which generative tools may save time at the margins, but also erode learning, judgment, and trust in the classroom.

Claire Vo races through Anthropic’s Code with Claude announcements, from scheduled routines in Claude Code to managed-agent features like outcome-based rubrics, multi-agent teams, and experimental memory tools. The conversation stays grounded in product work, sketching practical uses for newsletters, PRDs, and the growing appeal of agent systems that can grade, iterate, and remember.

A once-clunky coding tool is recast here as a desktop command center for modern office work, with hosts arguing that the real contest in AI is over agent interfaces that sit between workers and their apps. The conversation moves from theory to practice, showing how Codex and similar tools now draft emails, triage workflows, assemble strategy docs, and turn fragmented digital labor into something closer to managed delegation.

ZenBusiness CEO John Kim lays out an internal AI marketplace where employees post “quests,” share reusable skills, and compete on token-consumption leaderboards to turn curiosity into company-wide adoption. The conversation argues that AI works best not as a mandate to move faster, but as a way for marketers, operators, and leaders to build joyful, bespoke tools that would never survive a normal roadmap.

Simon Willison joins Joe Russo for a grounded assessment of AI coding tools as they move from autocomplete novelty to reliable engineering partners, reshaping review, mentorship, pricing and the pace of software work. Their conversation treats software development as a stubbornly human discipline even as models become powerful amplifiers of experience and increasingly capable collaborators.

MG Siegler and Alex Kantrowitz size up Apple’s unusually modest AI spending against the rest of Big Tech’s infrastructure binge, arguing that the company may be either prudently waiting out a commoditized market or dangerously ceding the future to rivals. They also trace the loosening Microsoft-OpenAI alliance and the murky reality of Stargate, where grand promises of owned infrastructure are giving way to outsourced capacity and financial improvisation.

A live conversation with Maven VP of product Rishin Banker examines how AI and new tools are collapsing traditional product, design, and engineering boundaries at a lean startup. The discussion traces what happens when more people can build, how teams reorganize into smaller pods, and where speed starts to create new handoff and governance problems.

A Stripe design leader describes building an internal prototyping stack that lets designers, PMs, and researchers generate realistic product demos in the browser, complete with live data, variants, and design-system guardrails. The conversation tracks how AI is collapsing the distance between mockup and product, turning static reviews into clickable demos and shifting collaboration from staffing debates to the work itself.

Uber CEO Dara Khosrowshahi sketches a company trying to turn its ride-hailing app into a broader travel and services platform, while juggling the trade-offs of hotel booking, personal shopping, AI tools and autonomous vehicles. He argues that Uber’s edge lies not in flashy demos but in handling the messy, probabilistic real world where reservations fail, drivers cancel and logistics have to keep working.

Max Schoening, Notion’s head of product, argues that AI is making the first draft of software nearly free while shifting the real differentiator to agency, taste and the discipline to make something obviously good. The conversation ranges from designers coding in terminals to the persistence of SaaS, all under a larger claim that the people who thrive will be the ones who act as if the world is changeable.

A jumble of AI signals comes into focus as OpenAI’s user growth cools, cloud providers post blockbuster numbers, and the industry lurches from consumer hype toward enterprise agents. Alex Kantrowitz and Ranjan Roy weigh whether generative AI is becoming ordinary infrastructure, a shaky consumer business, or simply a market still too expensive and too euphoric to understand clearly.

The Vergecast turns Elon Musk’s OpenAI trial testimony into a case study in how power evaporates under cross-examination, then widens into a sharper critique of AI’s shaky consumer appeal and the tech industry’s habit of mistaking compulsion for affection. Along the way, the hosts tour Microsoft and OpenAI’s retreat from the AGI fantasy, Brendan Carr’s latest FCC overreach, and a pile of gadgets that may or may not deserve Nilay Patel’s attention.

Ira Glass makes a funny, unexpectedly candid case against commencement speeches as a doomed genre, then explains why family history, old loyalties and one memorable Goucher dorm room changed his mind. The conversation sets up a 2012 graduation address shaped by skepticism, sentiment and a story about his grandmother meeting Adolf Hitler.

A veteran product leader argues that the best chief product officers set direction, understand customers deeply and keep reallocating resources as technology and markets shift. The conversation ranges from navigating founder dynamics and office politics to making riskier AI bets, protecting time to think and building products around outcomes rather than buttons.

Nilay Patel and Nick Statt turn Decoder’s own mailbag into a defense of adversarial tech journalism, using the backlash to a bruising Superhuman interview to argue for tougher scrutiny of AI hype and executive spin. The conversation draws a sharp line between enterprise use cases that may be real and consumer products that still feel coercive, unconvincing, and culturally corrosive.